MERGE-via-pending-tracking-hist-MERGE-via-stable-tracking-MERGE-via-mokopatches-tracking-fix-stray-endmenu-patch-1232632040-1232632141

pending-tracking-hist top was MERGE-via-stable-tracking-MERGE-via-mokopatches-tracking-fix-stray-endmenu-patch-1232632040-1232632141 / fdf777a63bcb59e0dfd78bfe2c6242e01f6d4eb9 ... parent commitmessage:
From: merge <null@invalid>
MERGE-via-stable-tracking-hist-MERGE-via-mokopatches-tracking-fix-stray-endmenu-patch-1232632040

stable-tracking-hist top was MERGE-via-mokopatches-tracking-fix-stray-endmenu-patch-1232632040 / 90463bfd2d5a3c8b52f6e6d71024a00e052b0ced ... parent commitmessage:
From: merge <null@invalid>
MERGE-via-mokopatches-tracking-hist-fix-stray-endmenu-patch

mokopatches-tracking-hist top was fix-stray-endmenu-patch / 3630e0be570de8057e7f8d2fe501ed353cdf34e6 ... parent commitmessage:
From: Andy Green <andy@openmoko.com>
fix-stray-endmenu.patch

Signed-off-by: Andy Green <andy@openmoko.com>

31 files changed, 1742 insertions, 1235 deletions

diff --git a/drivers/md/Makefile b/drivers/md/Makefile
index 1c615804ea7..72880b7e28d 100644
--- a/drivers/md/Makefile
+++ b/drivers/md/Makefile
@@ -3,9 +3,10 @@
 #
 
 dm-mod-objs	:= dm.o dm-table.o dm-target.o dm-linear.o dm-stripe.o \
-		   dm-ioctl.o dm-io.o dm-kcopyd.o
+		   dm-ioctl.o dm-io.o dm-kcopyd.o dm-sysfs.o
 dm-multipath-objs := dm-path-selector.o dm-mpath.o
-dm-snapshot-objs := dm-snap.o dm-exception-store.o
+dm-snapshot-objs := dm-snap.o dm-exception-store.o dm-snap-transient.o \
+		    dm-snap-persistent.o
 dm-mirror-objs	:= dm-raid1.o
 md-mod-objs     := md.o bitmap.o
 raid456-objs	:= raid5.o raid6algos.o raid6recov.o raid6tables.o \
diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c
index ac89a5deaca..71994376339 100644
--- a/drivers/md/bitmap.c
+++ b/drivers/md/bitmap.c
@@ -208,26 +208,30 @@ static void bitmap_checkfree(struct bitmap *bitmap, unsigned long page)
  */
 
 /* IO operations when bitmap is stored near all superblocks */
-static struct page *read_sb_page(mddev_t *mddev, long offset, unsigned long index)
+static struct page *read_sb_page(mddev_t *mddev, long offset,
+				 struct page *page,
+				 unsigned long index, int size)
 {
 	/* choose a good rdev and read the page from there */
 
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
-	struct page *page = alloc_page(GFP_KERNEL);
 	sector_t target;
 
 	if (!page)
+		page = alloc_page(GFP_KERNEL);
+	if (!page)
 		return ERR_PTR(-ENOMEM);
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		if (! test_bit(In_sync, &rdev->flags)
 		    || test_bit(Faulty, &rdev->flags))
 			continue;
 
 		target = rdev->sb_start + offset + index * (PAGE_SIZE/512);
 
-		if (sync_page_io(rdev->bdev, target, PAGE_SIZE, page, READ)) {
+		if (sync_page_io(rdev->bdev, target,
+				 roundup(size, bdev_hardsect_size(rdev->bdev)),
+				 page, READ)) {
 			page->index = index;
 			attach_page_buffers(page, NULL); /* so that free_buffer will
 							  * quietly no-op */
@@ -544,7 +548,9 @@ static int bitmap_read_sb(struct bitmap *bitmap)
 
 		bitmap->sb_page = read_page(bitmap->file, 0, bitmap, bytes);
 	} else {
-		bitmap->sb_page = read_sb_page(bitmap->mddev, bitmap->offset, 0);
+		bitmap->sb_page = read_sb_page(bitmap->mddev, bitmap->offset,
+					       NULL,
+					       0, sizeof(bitmap_super_t));
 	}
 	if (IS_ERR(bitmap->sb_page)) {
 		err = PTR_ERR(bitmap->sb_page);
@@ -957,11 +963,18 @@ static int bitmap_init_from_disk(struct bitmap *bitmap, sector_t start)
 				 */
 				page = bitmap->sb_page;
 				offset = sizeof(bitmap_super_t);
+				if (!file)
+					read_sb_page(bitmap->mddev,
+						     bitmap->offset,
+						     page,
+						     index, count);
 			} else if (file) {
 				page = read_page(file, index, bitmap, count);
 				offset = 0;
 			} else {
-				page = read_sb_page(bitmap->mddev, bitmap->offset, index);
+				page = read_sb_page(bitmap->mddev, bitmap->offset,
+						    NULL,
+						    index, count);
 				offset = 0;
 			}
 			if (IS_ERR(page)) { /* read error */
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index ce26c84af06..35bda49796f 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -1060,7 +1060,7 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 		goto bad_page_pool;
 	}
 
-	cc->bs = bioset_create(MIN_IOS, MIN_IOS);
+	cc->bs = bioset_create(MIN_IOS, 0);
 	if (!cc->bs) {
 		ti->error = "Cannot allocate crypt bioset";
 		goto bad_bs;
@@ -1322,11 +1322,7 @@ static int __init dm_crypt_init(void)
 
 static void __exit dm_crypt_exit(void)
 {
-	int r = dm_unregister_target(&crypt_target);
-
-	if (r < 0)
-		DMERR("unregister failed %d", r);
-
+	dm_unregister_target(&crypt_target);
 	kmem_cache_destroy(_crypt_io_pool);
 }
 
diff --git a/drivers/md/dm-delay.c b/drivers/md/dm-delay.c
index 848b381f117..59ee1b015d2 100644
--- a/drivers/md/dm-delay.c
+++ b/drivers/md/dm-delay.c
@@ -364,11 +364,7 @@ bad_queue:
 
 static void __exit dm_delay_exit(void)
 {
-	int r = dm_unregister_target(&delay_target);
-
-	if (r < 0)
-		DMERR("unregister failed %d", r);
-
+	dm_unregister_target(&delay_target);
 	kmem_cache_destroy(delayed_cache);
 	destroy_workqueue(kdelayd_wq);
 }
diff --git a/drivers/md/dm-exception-store.c b/drivers/md/dm-exception-store.c
index 01590f3e000..dccbfb0e010 100644
--- a/drivers/md/dm-exception-store.c
+++ b/drivers/md/dm-exception-store.c
@@ -1,756 +1,45 @@
 /*
- * dm-exception-store.c
- *
  * Copyright (C) 2001-2002 Sistina Software (UK) Limited.
- * Copyright (C) 2006 Red Hat GmbH
+ * Copyright (C) 2006-2008 Red Hat GmbH
  *
  * This file is released under the GPL.
  */
 
-#include "dm-snap.h"
+#include "dm-exception-store.h"
 
 #include <linux/mm.h>
 #include <linux/pagemap.h>
 #include <linux/vmalloc.h>
 #include <linux/slab.h>
-#include <linux/dm-io.h>
-#include <linux/dm-kcopyd.h>
-
-#define DM_MSG_PREFIX "snapshots"
-#define DM_CHUNK_SIZE_DEFAULT_SECTORS 32	/* 16KB */
-
-/*-----------------------------------------------------------------
- * Persistent snapshots, by persistent we mean that the snapshot
- * will survive a reboot.
- *---------------------------------------------------------------*/
-
-/*
- * We need to store a record of which parts of the origin have
- * been copied to the snapshot device.  The snapshot code
- * requires that we copy exception chunks to chunk aligned areas
- * of the COW store.  It makes sense therefore, to store the
- * metadata in chunk size blocks.
- *
- * There is no backward or forward compatibility implemented,
- * snapshots with different disk versions than the kernel will
- * not be usable.  It is expected that "lvcreate" will blank out
- * the start of a fresh COW device before calling the snapshot
- * constructor.
- *
- * The first chunk of the COW device just contains the header.
- * After this there is a chunk filled with exception metadata,
- * followed by as many exception chunks as can fit in the
- * metadata areas.
- *
- * All on disk structures are in little-endian format.  The end
- * of the exceptions info is indicated by an exception with a
- * new_chunk of 0, which is invalid since it would point to the
- * header chunk.
- */
-
-/*
- * Magic for persistent snapshots: "SnAp" - Feeble isn't it.
- */
-#define SNAP_MAGIC 0x70416e53
-
-/*
- * The on-disk version of the metadata.
- */
-#define SNAPSHOT_DISK_VERSION 1
-
-struct disk_header {
-	uint32_t magic;
-
-	/*
-	 * Is this snapshot valid.  There is no way of recovering
-	 * an invalid snapshot.
-	 */
-	uint32_t valid;
-
-	/*
-	 * Simple, incrementing version. no backward
-	 * compatibility.
-	 */
-	uint32_t version;
-
-	/* In sectors */
-	uint32_t chunk_size;
-};
-
-struct disk_exception {
-	uint64_t old_chunk;
-	uint64_t new_chunk;
-};
-
-struct commit_callback {
-	void (*callback)(void *, int success);
-	void *context;
-};
-
-/*
- * The top level structure for a persistent exception store.
- */
-struct pstore {
-	struct dm_snapshot *snap;	/* up pointer to my snapshot */
-	int version;
-	int valid;
-	uint32_t exceptions_per_area;
-
-	/*
-	 * Now that we have an asynchronous kcopyd there is no
-	 * need for large chunk sizes, so it wont hurt to have a
-	 * whole chunks worth of metadata in memory at once.
-	 */
-	void *area;
-
-	/*
-	 * An area of zeros used to clear the next area.
-	 */
-	void *zero_area;
-
-	/*
-	 * Used to keep track of which metadata area the data in
-	 * 'chunk' refers to.
-	 */
-	chunk_t current_area;
-
-	/*
-	 * The next free chunk for an exception.
-	 */
-	chunk_t next_free;
-
-	/*
-	 * The index of next free exception in the current
-	 * metadata area.
-	 */
-	uint32_t current_committed;
-
-	atomic_t pending_count;
-	uint32_t callback_count;
-	struct commit_callback *callbacks;
-	struct dm_io_client *io_client;
-
-	struct workqueue_struct *metadata_wq;
-};
-
-static unsigned sectors_to_pages(unsigned sectors)
-{
-	return DIV_ROUND_UP(sectors, PAGE_SIZE >> 9);
-}
-
-static int alloc_area(struct pstore *ps)
-{
-	int r = -ENOMEM;
-	size_t len;
-
-	len = ps->snap->chunk_size << SECTOR_SHIFT;
-
-	/*
-	 * Allocate the chunk_size block of memory that will hold
-	 * a single metadata area.
-	 */
-	ps->area = vmalloc(len);
-	if (!ps->area)
-		return r;
-
-	ps->zero_area = vmalloc(len);
-	if (!ps->zero_area) {
-		vfree(ps->area);
-		return r;
-	}
-	memset(ps->zero_area, 0, len);
-
-	return 0;
-}
-
-static void free_area(struct pstore *ps)
-{
-	vfree(ps->area);
-	ps->area = NULL;
-	vfree(ps->zero_area);
-	ps->zero_area = NULL;
-}
-
-struct mdata_req {
-	struct dm_io_region *where;
-	struct dm_io_request *io_req;
-	struct work_struct work;
-	int result;
-};
-
-static void do_metadata(struct work_struct *work)
-{
-	struct mdata_req *req = container_of(work, struct mdata_req, work);
-
-	req->result = dm_io(req->io_req, 1, req->where, NULL);
-}
-
-/*
- * Read or write a chunk aligned and sized block of data from a device.
- */
-static int chunk_io(struct pstore *ps, chunk_t chunk, int rw, int metadata)
-{
-	struct dm_io_region where = {
-		.bdev = ps->snap->cow->bdev,
-		.sector = ps->snap->chunk_size * chunk,
-		.count = ps->snap->chunk_size,
-	};
-	struct dm_io_request io_req = {
-		.bi_rw = rw,
-		.mem.type = DM_IO_VMA,
-		.mem.ptr.vma = ps->area,
-		.client = ps->io_client,
-		.notify.fn = NULL,
-	};
-	struct mdata_req req;
-
-	if (!metadata)
-		return dm_io(&io_req, 1, &where, NULL);
-
-	req.where = &where;
-	req.io_req = &io_req;
-
-	/*
-	 * Issue the synchronous I/O from a different thread
-	 * to avoid generic_make_request recursion.
-	 */
-	INIT_WORK(&req.work, do_metadata);
-	queue_work(ps->metadata_wq, &req.work);
-	flush_workqueue(ps->metadata_wq);
-
-	return req.result;
-}
-
-/*
- * Convert a metadata area index to a chunk index.
- */
-static chunk_t area_location(struct pstore *ps, chunk_t area)
-{
-	return 1 + ((ps->exceptions_per_area + 1) * area);
-}
-
-/*
- * Read or write a metadata area.  Remembering to skip the first
- * chunk which holds the header.
- */
-static int area_io(struct pstore *ps, int rw)
-{
-	int r;
-	chunk_t chunk;
-
-	chunk = area_location(ps, ps->current_area);
-
-	r = chunk_io(ps, chunk, rw, 0);
-	if (r)
-		return r;
-
-	return 0;
-}
-
-static void zero_memory_area(struct pstore *ps)
-{
-	memset(ps->area, 0, ps->snap->chunk_size << SECTOR_SHIFT);
-}
-
-static int zero_disk_area(struct pstore *ps, chunk_t area)
-{
-	struct dm_io_region where = {
-		.bdev = ps->snap->cow->bdev,
-		.sector = ps->snap->chunk_size * area_location(ps, area),
-		.count = ps->snap->chunk_size,
-	};
-	struct dm_io_request io_req = {
-		.bi_rw = WRITE,
-		.mem.type = DM_IO_VMA,
-		.mem.ptr.vma = ps->zero_area,
-		.client = ps->io_client,
-		.notify.fn = NULL,
-	};
-
-	return dm_io(&io_req, 1, &where, NULL);
-}
-
-static int read_header(struct pstore *ps, int *new_snapshot)
-{
-	int r;
-	struct disk_header *dh;
-	chunk_t chunk_size;
-	int chunk_size_supplied = 1;
-
-	/*
-	 * Use default chunk size (or hardsect_size, if larger) if none supplied
-	 */
-	if (!ps->snap->chunk_size) {
-        	ps->snap->chunk_size = max(DM_CHUNK_SIZE_DEFAULT_SECTORS,
-		    bdev_hardsect_size(ps->snap->cow->bdev) >> 9);
-		ps->snap->chunk_mask = ps->snap->chunk_size - 1;
-		ps->snap->chunk_shift = ffs(ps->snap->chunk_size) - 1;
-		chunk_size_supplied = 0;
-	}
-
-	ps->io_client = dm_io_client_create(sectors_to_pages(ps->snap->
-							     chunk_size));
-	if (IS_ERR(ps->io_client))
-		return PTR_ERR(ps->io_client);
-
-	r = alloc_area(ps);
-	if (r)
-		return r;
-
-	r = chunk_io(ps, 0, READ, 1);
-	if (r)
-		goto bad;
-
-	dh = (struct disk_header *) ps->area;
-
-	if (le32_to_cpu(dh->magic) == 0) {
-		*new_snapshot = 1;
-		return 0;
-	}
-
-	if (le32_to_cpu(dh->magic) != SNAP_MAGIC) {
-		DMWARN("Invalid or corrupt snapshot");
-		r = -ENXIO;
-		goto bad;
-	}
-
-	*new_snapshot = 0;
-	ps->valid = le32_to_cpu(dh->valid);
-	ps->version = le32_to_cpu(dh->version);
-	chunk_size = le32_to_cpu(dh->chunk_size);
-
-	if (!chunk_size_supplied || ps->snap->chunk_size == chunk_size)
-		return 0;
-
-	DMWARN("chunk size %llu in device metadata overrides "
-	       "table chunk size of %llu.",
-	       (unsigned long long)chunk_size,
-	       (unsigned long long)ps->snap->chunk_size);
-
-	/* We had a bogus chunk_size. Fix stuff up. */
-	free_area(ps);
-
-	ps->snap->chunk_size = chunk_size;
-	ps->snap->chunk_mask = chunk_size - 1;
-	ps->snap->chunk_shift = ffs(chunk_size) - 1;
-
-	r = dm_io_client_resize(sectors_to_pages(ps->snap->chunk_size),
-				ps->io_client);
-	if (r)
-		return r;
-
-	r = alloc_area(ps);
-	return r;
-
-bad:
-	free_area(ps);
-	return r;
-}
-
-static int write_header(struct pstore *ps)
-{
-	struct disk_header *dh;
-
-	memset(ps->area, 0, ps->snap->chunk_size << SECTOR_SHIFT);
-
-	dh = (struct disk_header *) ps->area;
-	dh->magic = cpu_to_le32(SNAP_MAGIC);
-	dh->valid = cpu_to_le32(ps->valid);
-	dh->version = cpu_to_le32(ps->version);
-	dh->chunk_size = cpu_to_le32(ps->snap->chunk_size);
-
-	return chunk_io(ps, 0, WRITE, 1);
-}
-
-/*
- * Access functions for the disk exceptions, these do the endian conversions.
- */
-static struct disk_exception *get_exception(struct pstore *ps, uint32_t index)
-{
-	BUG_ON(index >= ps->exceptions_per_area);
-
-	return ((struct disk_exception *) ps->area) + index;
-}
 
-static void read_exception(struct pstore *ps,
-			   uint32_t index, struct disk_exception *result)
-{
-	struct disk_exception *e = get_exception(ps, index);
-
-	/* copy it */
-	result->old_chunk = le64_to_cpu(e->old_chunk);
-	result->new_chunk = le64_to_cpu(e->new_chunk);
-}
-
-static void write_exception(struct pstore *ps,
-			    uint32_t index, struct disk_exception *de)
-{
-	struct disk_exception *e = get_exception(ps, index);
-
-	/* copy it */
-	e->old_chunk = cpu_to_le64(de->old_chunk);
-	e->new_chunk = cpu_to_le64(de->new_chunk);
-}
+#define DM_MSG_PREFIX "snapshot exception stores"
 
-/*
- * Registers the exceptions that are present in the current area.
- * 'full' is filled in to indicate if the area has been
- * filled.
- */
-static int insert_exceptions(struct pstore *ps, int *full)
+int dm_exception_store_init(void)
 {
 	int r;
-	unsigned int i;
-	struct disk_exception de;
-
-	/* presume the area is full */
-	*full = 1;
-
-	for (i = 0; i < ps->exceptions_per_area; i++) {
-		read_exception(ps, i, &de);
-
-		/*
-		 * If the new_chunk is pointing at the start of
-		 * the COW device, where the first metadata area
-		 * is we know that we've hit the end of the
-		 * exceptions.  Therefore the area is not full.
-		 */
-		if (de.new_chunk == 0LL) {
-			ps->current_committed = i;
-			*full = 0;
-			break;
-		}
-
-		/*
-		 * Keep track of the start of the free chunks.
-		 */
-		if (ps->next_free <= de.new_chunk)
-			ps->next_free = de.new_chunk + 1;
-
-		/*
-		 * Otherwise we add the exception to the snapshot.
-		 */
-		r = dm_add_exception(ps->snap, de.old_chunk, de.new_chunk);
-		if (r)
-			return r;
-	}
-
-	return 0;
-}
-
-static int read_exceptions(struct pstore *ps)
-{
-	int r, full = 1;
-
-	/*
-	 * Keeping reading chunks and inserting exceptions until
-	 * we find a partially full area.
-	 */
-	for (ps->current_area = 0; full; ps->current_area++) {
-		r = area_io(ps, READ);
-		if (r)
-			return r;
 
-		r = insert_exceptions(ps, &full);
-		if (r)
-			return r;
+	r = dm_transient_snapshot_init();
+	if (r) {
+		DMERR("Unable to register transient exception store type.");
+		goto transient_fail;
 	}
 
-	ps->current_area--;
-
-	return 0;
-}
-
-static struct pstore *get_info(struct exception_store *store)
-{
-	return (struct pstore *) store->context;
-}
-
-static void persistent_fraction_full(struct exception_store *store,
-				     sector_t *numerator, sector_t *denominator)
-{
-	*numerator = get_info(store)->next_free * store->snap->chunk_size;
-	*denominator = get_dev_size(store->snap->cow->bdev);
-}
-
-static void persistent_destroy(struct exception_store *store)
-{
-	struct pstore *ps = get_info(store);
-
-	destroy_workqueue(ps->metadata_wq);
-	dm_io_client_destroy(ps->io_client);
-	vfree(ps->callbacks);
-	free_area(ps);
-	kfree(ps);
-}
-
-static int persistent_read_metadata(struct exception_store *store)
-{
-	int r, uninitialized_var(new_snapshot);
-	struct pstore *ps = get_info(store);
-
-	/*
-	 * Read the snapshot header.
-	 */
-	r = read_header(ps, &new_snapshot);
-	if (r)
-		return r;
-
-	/*
-	 * Now we know correct chunk_size, complete the initialisation.
-	 */
-	ps->exceptions_per_area = (ps->snap->chunk_size << SECTOR_SHIFT) /
-				  sizeof(struct disk_exception);
-	ps->callbacks = dm_vcalloc(ps->exceptions_per_area,
-			sizeof(*ps->callbacks));
-	if (!ps->callbacks)
-		return -ENOMEM;
-
-	/*
-	 * Do we need to setup a new snapshot ?
-	 */
-	if (new_snapshot) {
-		r = write_header(ps);
-		if (r) {
-			DMWARN("write_header failed");
-			return r;
-		}
-
-		ps->current_area = 0;
-		zero_memory_area(ps);
-		r = zero_disk_area(ps, 0);
-		if (r) {
-			DMWARN("zero_disk_area(0) failed");
-			return r;
-		}
-	} else {
-		/*
-		 * Sanity checks.
-		 */
-		if (ps->version != SNAPSHOT_DISK_VERSION) {
-			DMWARN("unable to handle snapshot disk version %d",
-			       ps->version);
-			return -EINVAL;
-		}
-
-		/*
-		 * Metadata are valid, but snapshot is invalidated
-		 */
-		if (!ps->valid)
-			return 1;
-
-		/*
-		 * Read the metadata.
-		 */
-		r = read_exceptions(ps);
-		if (r)
-			return r;
+	r = dm_persistent_snapshot_init();
+	if (r) {
+		DMERR("Unable to register persistent exception store type");
+		goto persistent_fail;
 	}
 
 	return 0;
-}
-
-static int persistent_prepare(struct exception_store *store,
-			      struct dm_snap_exception *e)
-{
-	struct pstore *ps = get_info(store);
-	uint32_t stride;
-	chunk_t next_free;
-	sector_t size = get_dev_size(store->snap->cow->bdev);
-
-	/* Is there enough room ? */
-	if (size < ((ps->next_free + 1) * store->snap->chunk_size))
-		return -ENOSPC;
 
-	e->new_chunk = ps->next_free;
-
-	/*
-	 * Move onto the next free pending, making sure to take
-	 * into account the location of the metadata chunks.
-	 */
-	stride = (ps->exceptions_per_area + 1);
-	next_free = ++ps->next_free;
-	if (sector_div(next_free, stride) == 1)
-		ps->next_free++;
-
-	atomic_inc(&ps->pending_count);
-	return 0;
-}
-
-static void persistent_commit(struct exception_store *store,
-			      struct dm_snap_exception *e,
-			      void (*callback) (void *, int success),
-			      void *callback_context)
-{
-	unsigned int i;
-	struct pstore *ps = get_info(store);
-	struct disk_exception de;
-	struct commit_callback *cb;
-
-	de.old_chunk = e->old_chunk;
-	de.new_chunk = e->new_chunk;
-	write_exception(ps, ps->current_committed++, &de);
-
-	/*
-	 * Add the callback to the back of the array.  This code
-	 * is the only place where the callback array is
-	 * manipulated, and we know that it will never be called
-	 * multiple times concurrently.
-	 */
-	cb = ps->callbacks + ps->callback_count++;
-	cb->callback = callback;
-	cb->context = callback_context;
-
-	/*
-	 * If there are exceptions in flight and we have not yet
-	 * filled this metadata area there's nothing more to do.
-	 */
-	if (!atomic_dec_and_test(&ps->pending_count) &&
-	    (ps->current_committed != ps->exceptions_per_area))
-		return;
-
-	/*
-	 * If we completely filled the current area, then wipe the next one.
-	 */
-	if ((ps->current_committed == ps->exceptions_per_area) &&
-	     zero_disk_area(ps, ps->current_area + 1))
-		ps->valid = 0;
-
-	/*
-	 * Commit exceptions to disk.
-	 */
-	if (ps->valid && area_io(ps, WRITE))
-		ps->valid = 0;
-
-	/*
-	 * Advance to the next area if this one is full.
-	 */
-	if (ps->current_committed == ps->exceptions_per_area) {
-		ps->current_committed = 0;
-		ps->current_area++;
-		zero_memory_area(ps);
-	}
-
-	for (i = 0; i < ps->callback_count; i++) {
-		cb = ps->callbacks + i;
-		cb->callback(cb->context, ps->valid);
-	}
-
-	ps->callback_count = 0;
-}
-
-static void persistent_drop(struct exception_store *store)
-{
-	struct pstore *ps = get_info(store);
-
-	ps->valid = 0;
-	if (write_header(ps))
-		DMWARN("write header failed");
-}
-
-int dm_create_persistent(struct exception_store *store)
-{
-	struct pstore *ps;
-
-	/* allocate the pstore */
-	ps = kmalloc(sizeof(*ps), GFP_KERNEL);
-	if (!ps)
-		return -ENOMEM;
-
-	ps->snap = store->snap;
-	ps->valid = 1;
-	ps->version = SNAPSHOT_DISK_VERSION;
-	ps->area = NULL;
-	ps->next_free = 2;	/* skipping the header and first area */
-	ps->current_committed = 0;
-
-	ps->callback_count = 0;
-	atomic_set(&ps->pending_count, 0);
-	ps->callbacks = NULL;
-
-	ps->metadata_wq = create_singlethread_workqueue("ksnaphd");
-	if (!ps->metadata_wq) {
-		kfree(ps);
-		DMERR("couldn't start header metadata update thread");
-		return -ENOMEM;
-	}
-
-	store->destroy = persistent_destroy;
-	store->read_metadata = persistent_read_metadata;
-	store->prepare_exception = persistent_prepare;
-	store->commit_exception = persistent_commit;
-	store->drop_snapshot = persistent_drop;
-	store->fraction_full = persistent_fraction_full;
-	store->context = ps;
-
-	return 0;
-}
-
-/*-----------------------------------------------------------------
- * Implementation of the store for non-persistent snapshots.
- *---------------------------------------------------------------*/
-struct transient_c {
-	sector_t next_free;
-};
-
-static void transient_destroy(struct exception_store *store)
-{
-	kfree(store->context);
-}
-
-static int transient_read_metadata(struct exception_store *store)
-{
-	return 0;
-}
-
-static int transient_prepare(struct exception_store *store,
-			     struct dm_snap_exception *e)
-{
-	struct transient_c *tc = (struct transient_c *) store->context;
-	sector_t size = get_dev_size(store->snap->cow->bdev);
-
-	if (size < (tc->next_free + store->snap->chunk_size))
-		return -1;
-
-	e->new_chunk = sector_to_chunk(store->snap, tc->next_free);
-	tc->next_free += store->snap->chunk_size;
-
-	return 0;
-}
-
-static void transient_commit(struct exception_store *store,
-			     struct dm_snap_exception *e,
-			     void (*callback) (void *, int success),
-			     void *callback_context)
-{
-	/* Just succeed */
-	callback(callback_context, 1);
-}
-
-static void transient_fraction_full(struct exception_store *store,
-				    sector_t *numerator, sector_t *denominator)
-{
-	*numerator = ((struct transient_c *) store->context)->next_free;
-	*denominator = get_dev_size(store->snap->cow->bdev);
+persistent_fail:
+	dm_persistent_snapshot_exit();
+transient_fail:
+	return r;
 }
 
-int dm_create_transient(struct exception_store *store)
+void dm_exception_store_exit(void)
 {
-	struct transient_c *tc;
-
-	store->destroy = transient_destroy;
-	store->read_metadata = transient_read_metadata;
-	store->prepare_exception = transient_prepare;
-	store->commit_exception = transient_commit;
-	store->drop_snapshot = NULL;
-	store->fraction_full = transient_fraction_full;
-
-	tc = kmalloc(sizeof(struct transient_c), GFP_KERNEL);
-	if (!tc)
-		return -ENOMEM;
-
-	tc->next_free = 0;
-	store->context = tc;
-
-	return 0;
+	dm_persistent_snapshot_exit();
+	dm_transient_snapshot_exit();
 }
diff --git a/drivers/md/dm-exception-store.h b/drivers/md/dm-exception-store.h
new file mode 100644
index 00000000000..bb9f33d5daa
--- /dev/null
+++ b/drivers/md/dm-exception-store.h
@@ -0,0 +1,148 @@
+/*
+ * Copyright (C) 2001-2002 Sistina Software (UK) Limited.
+ * Copyright (C) 2008 Red Hat, Inc. All rights reserved.
+ *
+ * Device-mapper snapshot exception store.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LINUX_DM_EXCEPTION_STORE
+#define _LINUX_DM_EXCEPTION_STORE
+
+#include <linux/blkdev.h>
+#include <linux/device-mapper.h>
+
+/*
+ * The snapshot code deals with largish chunks of the disk at a
+ * time. Typically 32k - 512k.
+ */
+typedef sector_t chunk_t;
+
+/*
+ * An exception is used where an old chunk of data has been
+ * replaced by a new one.
+ * If chunk_t is 64 bits in size, the top 8 bits of new_chunk hold the number
+ * of chunks that follow contiguously.  Remaining bits hold the number of the
+ * chunk within the device.
+ */
+struct dm_snap_exception {
+	struct list_head hash_list;
+
+	chunk_t old_chunk;
+	chunk_t new_chunk;
+};
+
+/*
+ * Abstraction to handle the meta/layout of exception stores (the
+ * COW device).
+ */
+struct dm_exception_store {
+	/*
+	 * Destroys this object when you've finished with it.
+	 */
+	void (*destroy) (struct dm_exception_store *store);
+
+	/*
+	 * The target shouldn't read the COW device until this is
+	 * called.  As exceptions are read from the COW, they are
+	 * reported back via the callback.
+	 */
+	int (*read_metadata) (struct dm_exception_store *store,
+			      int (*callback)(void *callback_context,
+					      chunk_t old, chunk_t new),
+			      void *callback_context);
+
+	/*
+	 * Find somewhere to store the next exception.
+	 */
+	int (*prepare_exception) (struct dm_exception_store *store,
+				  struct dm_snap_exception *e);
+
+	/*
+	 * Update the metadata with this exception.
+	 */
+	void (*commit_exception) (struct dm_exception_store *store,
+				  struct dm_snap_exception *e,
+				  void (*callback) (void *, int success),
+				  void *callback_context);
+
+	/*
+	 * The snapshot is invalid, note this in the metadata.
+	 */
+	void (*drop_snapshot) (struct dm_exception_store *store);
+
+	int (*status) (struct dm_exception_store *store, status_type_t status,
+		       char *result, unsigned int maxlen);
+
+	/*
+	 * Return how full the snapshot is.
+	 */
+	void (*fraction_full) (struct dm_exception_store *store,
+			       sector_t *numerator,
+			       sector_t *denominator);
+
+	struct dm_snapshot *snap;
+	void *context;
+};
+
+/*
+ * Funtions to manipulate consecutive chunks
+ */
+#  if defined(CONFIG_LBD) || (BITS_PER_LONG == 64)
+#    define DM_CHUNK_CONSECUTIVE_BITS 8
+#    define DM_CHUNK_NUMBER_BITS 56
+
+static inline chunk_t dm_chunk_number(chunk_t chunk)
+{
+	return chunk & (chunk_t)((1ULL << DM_CHUNK_NUMBER_BITS) - 1ULL);
+}
+
+static inline unsigned dm_consecutive_chunk_count(struct dm_snap_exception *e)
+{
+	return e->new_chunk >> DM_CHUNK_NUMBER_BITS;
+}
+
+static inline void dm_consecutive_chunk_count_inc(struct dm_snap_exception *e)
+{
+	e->new_chunk += (1ULL << DM_CHUNK_NUMBER_BITS);
+
+	BUG_ON(!dm_consecutive_chunk_count(e));
+}
+
+#  else
+#    define DM_CHUNK_CONSECUTIVE_BITS 0
+
+static inline chunk_t dm_chunk_number(chunk_t chunk)
+{
+	return chunk;
+}
+
+static inline unsigned dm_consecutive_chunk_count(struct dm_snap_exception *e)
+{
+	return 0;
+}
+
+static inline void dm_consecutive_chunk_count_inc(struct dm_snap_exception *e)
+{
+}
+
+#  endif
+
+int dm_exception_store_init(void);
+void dm_exception_store_exit(void);
+
+/*
+ * Two exception store implementations.
+ */
+int dm_persistent_snapshot_init(void);
+void dm_persistent_snapshot_exit(void);
+
+int dm_transient_snapshot_init(void);
+void dm_transient_snapshot_exit(void);
+
+int dm_create_persistent(struct dm_exception_store *store);
+
+int dm_create_transient(struct dm_exception_store *store);
+
+#endif /* _LINUX_DM_EXCEPTION_STORE */
diff --git a/drivers/md/dm-io.c b/drivers/md/dm-io.c
index 2fd6d445063..a34338567a2 100644
--- a/drivers/md/dm-io.c
+++ b/drivers/md/dm-io.c
@@ -56,7 +56,7 @@ struct dm_io_client *dm_io_client_create(unsigned num_pages)
 	if (!client->pool)
 		goto bad;
 
-	client->bios = bioset_create(16, 16);
+	client->bios = bioset_create(16, 0);
 	if (!client->bios)
 		goto bad;
 
diff --git a/drivers/md/dm-ioctl.c b/drivers/md/dm-ioctl.c
index 777c948180f..54d0588fc1f 100644
--- a/drivers/md/dm-ioctl.c
+++ b/drivers/md/dm-ioctl.c
@@ -233,7 +233,7 @@ static void __hash_remove(struct hash_cell *hc)
 	}
 
 	if (hc->new_map)
-		dm_table_put(hc->new_map);
+		dm_table_destroy(hc->new_map);
 	dm_put(hc->md);
 	free_cell(hc);
 }
@@ -827,8 +827,8 @@ static int do_resume(struct dm_ioctl *param)
 
 		r = dm_swap_table(md, new_map);
 		if (r) {
+			dm_table_destroy(new_map);
 			dm_put(md);
-			dm_table_put(new_map);
 			return r;
 		}
 
@@ -836,8 +836,6 @@ static int do_resume(struct dm_ioctl *param)
 			set_disk_ro(dm_disk(md), 0);
 		else
 			set_disk_ro(dm_disk(md), 1);
-
-		dm_table_put(new_map);
 	}
 
 	if (dm_suspended(md))
@@ -1080,7 +1078,7 @@ static int table_load(struct dm_ioctl *param, size_t param_size)
 	}
 
 	if (hc->new_map)
-		dm_table_put(hc->new_map);
+		dm_table_destroy(hc->new_map);
 	hc->new_map = t;
 	up_write(&_hash_lock);
 
@@ -1109,7 +1107,7 @@ static int table_clear(struct dm_ioctl *param, size_t param_size)
 	}
 
 	if (hc->new_map) {
-		dm_table_put(hc->new_map);
+		dm_table_destroy(hc->new_map);
 		hc->new_map = NULL;
 	}
 
@@ -1550,8 +1548,10 @@ int dm_copy_name_and_uuid(struct mapped_device *md, char *name, char *uuid)
 		goto out;
 	}
 
-	strcpy(name, hc->name);
-	strcpy(uuid, hc->uuid ? : "");
+	if (name)
+		strcpy(name, hc->name);
+	if (uuid)
+		strcpy(uuid, hc->uuid ? : "");
 
 out:
 	up_read(&_hash_lock);
diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c
index 44042becad8..bfa107f59d9 100644
--- a/drivers/md/dm-linear.c
+++ b/drivers/md/dm-linear.c
@@ -142,6 +142,7 @@ static struct target_type linear_target = {
 	.status = linear_status,
 	.ioctl  = linear_ioctl,
 	.merge  = linear_merge,
+	.features = DM_TARGET_SUPPORTS_BARRIERS,
 };
 
 int __init dm_linear_init(void)
@@ -156,8 +157,5 @@ int __init dm_linear_init(void)
 
 void dm_linear_exit(void)
 {
-	int r = dm_unregister_target(&linear_target);
-
-	if (r < 0)
-		DMERR("unregister failed %d", r);
+	dm_unregister_target(&linear_target);
 }
diff --git a/drivers/md/dm-log.c b/drivers/md/dm-log.c
index a8c0fc79ca7..737961f275c 100644
--- a/drivers/md/dm-log.c
+++ b/drivers/md/dm-log.c
@@ -326,8 +326,6 @@ static void header_from_disk(struct log_header *core, struct log_header *disk)
 static int rw_header(struct log_c *lc, int rw)
 {
 	lc->io_req.bi_rw = rw;
-	lc->io_req.mem.ptr.vma = lc->disk_header;
-	lc->io_req.notify.fn = NULL;
 
 	return dm_io(&lc->io_req, 1, &lc->header_location, NULL);
 }
@@ -362,10 +360,15 @@ static int read_header(struct log_c *log)
 	return 0;
 }
 
-static inline int write_header(struct log_c *log)
+static int _check_region_size(struct dm_target *ti, uint32_t region_size)
 {
-	header_to_disk(&log->header, log->disk_header);
-	return rw_header(log, WRITE);
+	if (region_size < 2 || region_size > ti->len)
+		return 0;
+
+	if (!is_power_of_2(region_size))
+		return 0;
+
+	return 1;
 }
 
 /*----------------------------------------------------------------
@@ -403,8 +406,9 @@ static int create_log_context(struct dm_dirty_log *log, struct dm_target *ti,
 		}
 	}
 
-	if (sscanf(argv[0], "%u", &region_size) != 1) {
-		DMWARN("invalid region size string");
+	if (sscanf(argv[0], "%u", &region_size) != 1 ||
+	    !_check_region_size(ti, region_size)) {
+		DMWARN("invalid region size %s", argv[0]);
 		return -EINVAL;
 	}
 
@@ -453,8 +457,18 @@ static int create_log_context(struct dm_dirty_log *log, struct dm_target *ti,
 		 */
 		buf_size = dm_round_up((LOG_OFFSET << SECTOR_SHIFT) +
 				       bitset_size, ti->limits.hardsect_size);
+
+		if (buf_size > dev->bdev->bd_inode->i_size) {
+			DMWARN("log device %s too small: need %llu bytes",
+				dev->name, (unsigned long long)buf_size);
+			kfree(lc);
+			return -EINVAL;
+		}
+
 		lc->header_location.count = buf_size >> SECTOR_SHIFT;
+
 		lc->io_req.mem.type = DM_IO_VMA;
+		lc->io_req.notify.fn = NULL;
 		lc->io_req.client = dm_io_client_create(dm_div_up(buf_size,
 								   PAGE_SIZE));
 		if (IS_ERR(lc->io_req.client)) {
@@ -467,10 +481,12 @@ static int create_log_context(struct dm_dirty_log *log, struct dm_target *ti,
 		lc->disk_header = vmalloc(buf_size);
 		if (!lc->disk_header) {
 			DMWARN("couldn't allocate disk log buffer");
+			dm_io_client_destroy(lc->io_req.client);
 			kfree(lc);
 			return -ENOMEM;
 		}
 
+		lc->io_req.mem.ptr.vma = lc->disk_header;
 		lc->clean_bits = (void *)lc->disk_header +
 				 (LOG_OFFSET << SECTOR_SHIFT);
 	}
@@ -482,6 +498,8 @@ static int create_log_context(struct dm_dirty_log *log, struct dm_target *ti,
 		DMWARN("couldn't allocate sync bitset");
 		if (!dev)
 			vfree(lc->clean_bits);
+		else
+			dm_io_client_destroy(lc->io_req.client);
 		vfree(lc->disk_header);
 		kfree(lc);
 		return -ENOMEM;
@@ -495,6 +513,8 @@ static int create_log_context(struct dm_dirty_log *log, struct dm_target *ti,
 		vfree(lc->sync_bits);
 		if (!dev)
 			vfree(lc->clean_bits);
+		else
+			dm_io_client_destroy(lc->io_req.client);
 		vfree(lc->disk_header);
 		kfree(lc);
 		return -ENOMEM;
@@ -631,8 +651,10 @@ static int disk_resume(struct dm_dirty_log *log)
 	/* set the correct number of regions in the header */
 	lc->header.nr_regions = lc->region_count;
 
+	header_to_disk(&lc->header, lc->disk_header);
+
 	/* write the new header */
-	r = write_header(lc);
+	r = rw_header(lc, WRITE);
 	if (r) {
 		DMWARN("%s: Failed to write header on dirty region log device",
 		       lc->log_dev->name);
@@ -682,7 +704,7 @@ static int disk_flush(struct dm_dirty_log *log)
 	if (!lc->touched)
 		return 0;
 
-	r = write_header(lc);
+	r = rw_header(lc, WRITE);
 	if (r)
 		fail_log_device(lc);
 	else
diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 4840733cd90..095f77bf968 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -441,13 +441,13 @@ static void process_queued_ios(struct work_struct *work)
 		__choose_pgpath(m);
 
 	pgpath = m->current_pgpath;
-	m->pgpath_to_activate = m->current_pgpath;
 
 	if ((pgpath && !m->queue_io) ||
 	    (!pgpath && !m->queue_if_no_path))
 		must_queue = 0;
 
-	if (m->pg_init_required && !m->pg_init_in_progress) {
+	if (m->pg_init_required && !m->pg_init_in_progress && pgpath) {
+		m->pgpath_to_activate = pgpath;
 		m->pg_init_count++;
 		m->pg_init_required = 0;
 		m->pg_init_in_progress = 1;
@@ -708,6 +708,10 @@ static int parse_hw_handler(struct arg_set *as, struct multipath *m)
 		m->hw_handler_name = NULL;
 		return -EINVAL;
 	}
+
+	if (hw_argc > 1)
+		DMWARN("Ignoring user-specified arguments for "
+		       "hardware handler \"%s\"", m->hw_handler_name);
 	consume(as, hw_argc - 1);
 
 	return 0;
@@ -885,7 +889,7 @@ static int fail_path(struct pgpath *pgpath)
 	dm_path_uevent(DM_UEVENT_PATH_FAILED, m->ti,
 		      pgpath->path.dev->name, m->nr_valid_paths);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 	queue_work(kmultipathd, &pgpath->deactivate_path);
 
 out:
@@ -928,7 +932,7 @@ static int reinstate_path(struct pgpath *pgpath)
 	dm_path_uevent(DM_UEVENT_PATH_REINSTATED, m->ti,
 		      pgpath->path.dev->name, m->nr_valid_paths);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 
 out:
 	spin_unlock_irqrestore(&m->lock, flags);
@@ -972,7 +976,7 @@ static void bypass_pg(struct multipath *m, struct priority_group *pg,
 
 	spin_unlock_irqrestore(&m->lock, flags);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 }
 
 /*
@@ -1002,7 +1006,7 @@ static int switch_pg_num(struct multipath *m, const char *pgstr)
 	}
 	spin_unlock_irqrestore(&m->lock, flags);
 
-	queue_work(kmultipathd, &m->trigger_event);
+	schedule_work(&m->trigger_event);
 	return 0;
 }
 
@@ -1491,14 +1495,10 @@ static int __init dm_multipath_init(void)
 
 static void __exit dm_multipath_exit(void)
 {
-	int r;
-
 	destroy_workqueue(kmpath_handlerd);
 	destroy_workqueue(kmultipathd);
 
-	r = dm_unregister_target(&multipath_target);
-	if (r < 0)
-		DMERR("target unregister failed %d", r);
+	dm_unregister_target(&multipath_target);
 	kmem_cache_destroy(_mpio_cache);
 }
 
diff --git a/drivers/md/dm-raid1.c b/drivers/md/dm-raid1.c
index 9d7b53ed75b..4d6bc101962 100644
--- a/drivers/md/dm-raid1.c
+++ b/drivers/md/dm-raid1.c
@@ -197,9 +197,6 @@ static void fail_mirror(struct mirror *m, enum dm_raid1_error error_type)
 	struct mirror_set *ms = m->ms;
 	struct mirror *new;
 
-	if (!errors_handled(ms))
-		return;
-
 	/*
 	 * error_count is used for nothing more than a
 	 * simple way to tell if a device has encountered
@@ -210,6 +207,9 @@ static void fail_mirror(struct mirror *m, enum dm_raid1_error error_type)
 	if (test_and_set_bit(error_type, &m->error_type))
 		return;
 
+	if (!errors_handled(ms))
+		return;
+
 	if (m != get_default_mirror(ms))
 		goto out;
 
@@ -808,12 +808,6 @@ static void free_context(struct mirror_set *ms, struct dm_target *ti,
 	kfree(ms);
 }
 
-static inline int _check_region_size(struct dm_target *ti, uint32_t size)
-{
-	return !(size % (PAGE_SIZE >> 9) || !is_power_of_2(size) ||
-		 size > ti->len);
-}
-
 static int get_mirror(struct mirror_set *ms, struct dm_target *ti,
 		      unsigned int mirror, char **argv)
 {
@@ -872,12 +866,6 @@ static struct dm_dirty_log *create_dirty_log(struct dm_target *ti,
 		return NULL;
 	}
 
-	if (!_check_region_size(ti, dl->type->get_region_size(dl))) {
-		ti->error = "Invalid region size";
-		dm_dirty_log_destroy(dl);
-		return NULL;
-	}
-
 	return dl;
 }
 
@@ -1032,6 +1020,7 @@ static void mirror_dtr(struct dm_target *ti)
 
 	del_timer_sync(&ms->timer);
 	flush_workqueue(ms->kmirrord_wq);
+	flush_scheduled_work();
 	dm_kcopyd_client_destroy(ms->kcopyd_client);
 	destroy_workqueue(ms->kmirrord_wq);
 	free_context(ms, ti, ms->nr_mirrors);
@@ -1299,11 +1288,7 @@ static int __init dm_mirror_init(void)
 
 static void __exit dm_mirror_exit(void)
 {
-	int r;
-
-	r = dm_unregister_target(&mirror_target);
-	if (r < 0)
-		DMERR("unregister failed %d", r);
+	dm_unregister_target(&mirror_target);
 }
 
 /* Module hooks */
diff --git a/drivers/md/dm-snap-persistent.c b/drivers/md/dm-snap-persistent.c
new file mode 100644
index 00000000000..936b34e0959
--- /dev/null
+++ b/drivers/md/dm-snap-persistent.c
@@ -0,0 +1,704 @@
+/*
+ * Copyright (C) 2001-2002 Sistina Software (UK) Limited.
+ * Copyright (C) 2006-2008 Red Hat GmbH
+ *
+ * This file is released under the GPL.
+ */
+
+#include "dm-exception-store.h"
+#include "dm-snap.h"
+
+#include <linux/mm.h>
+#include <linux/pagemap.h>
+#include <linux/vmalloc.h>
+#include <linux/slab.h>
+#include <linux/dm-io.h>
+
+#define DM_MSG_PREFIX "persistent snapshot"
+#define DM_CHUNK_SIZE_DEFAULT_SECTORS 32	/* 16KB */
+
+/*-----------------------------------------------------------------
+ * Persistent snapshots, by persistent we mean that the snapshot
+ * will survive a reboot.
+ *---------------------------------------------------------------*/
+
+/*
+ * We need to store a record of which parts of the origin have
+ * been copied to the snapshot device.  The snapshot code
+ * requires that we copy exception chunks to chunk aligned areas
+ * of the COW store.  It makes sense therefore, to store the
+ * metadata in chunk size blocks.
+ *
+ * There is no backward or forward compatibility implemented,
+ * snapshots with different disk versions than the kernel will
+ * not be usable.  It is expected that "lvcreate" will blank out
+ * the start of a fresh COW device before calling the snapshot
+ * constructor.
+ *
+ * The first chunk of the COW device just contains the header.
+ * After this there is a chunk filled with exception metadata,
+ * followed by as many exception chunks as can fit in the
+ * metadata areas.
+ *
+ * All on disk structures are in little-endian format.  The end
+ * of the exceptions info is indicated by an exception with a
+ * new_chunk of 0, which is invalid since it would point to the
+ * header chunk.
+ */
+
+/*
+ * Magic for persistent snapshots: "SnAp" - Feeble isn't it.
+ */
+#define SNAP_MAGIC 0x70416e53
+
+/*
+ * The on-disk version of the metadata.
+ */
+#define SNAPSHOT_DISK_VERSION 1
+
+struct disk_header {
+	uint32_t magic;
+
+	/*
+	 * Is this snapshot valid.  There is no way of recovering
+	 * an invalid snapshot.
+	 */
+	uint32_t valid;
+
+	/*
+	 * Simple, incrementing version. no backward
+	 * compatibility.
+	 */
+	uint32_t version;
+
+	/* In sectors */
+	uint32_t chunk_size;
+};
+
+struct disk_exception {
+	uint64_t old_chunk;
+	uint64_t new_chunk;
+};
+
+struct commit_callback {
+	void (*callback)(void *, int success);
+	void *context;
+};
+
+/*
+ * The top level structure for a persistent exception store.
+ */
+struct pstore {
+	struct dm_snapshot *snap;	/* up pointer to my snapshot */
+	int version;
+	int valid;
+	uint32_t exceptions_per_area;
+
+	/*
+	 * Now that we have an asynchronous kcopyd there is no
+	 * need for large chunk sizes, so it wont hurt to have a
+	 * whole chunks worth of metadata in memory at once.
+	 */
+	void *area;
+
+	/*
+	 * An area of zeros used to clear the next area.
+	 */
+	void *zero_area;
+
+	/*
+	 * Used to keep track of which metadata area the data in
+	 * 'chunk' refers to.
+	 */
+	chunk_t current_area;
+
+	/*
+	 * The next free chunk for an exception.
+	 */
+	chunk_t next_free;
+
+	/*
+	 * The index of next free exception in the current
+	 * metadata area.
+	 */
+	uint32_t current_committed;
+
+	atomic_t pending_count;
+	uint32_t callback_count;
+	struct commit_callback *callbacks;
+	struct dm_io_client *io_client;
+
+	struct workqueue_struct *metadata_wq;
+};
+
+static unsigned sectors_to_pages(unsigned sectors)
+{
+	return DIV_ROUND_UP(sectors, PAGE_SIZE >> 9);
+}
+
+static int alloc_area(struct pstore *ps)
+{
+	int r = -ENOMEM;
+	size_t len;
+
+	len = ps->snap->chunk_size << SECTOR_SHIFT;
+
+	/*
+	 * Allocate the chunk_size block of memory that will hold
+	 * a single metadata area.
+	 */
+	ps->area = vmalloc(len);
+	if (!ps->area)
+		return r;
+
+	ps->zero_area = vmalloc(len);
+	if (!ps->zero_area) {
+		vfree(ps->area);
+		return r;
+	}
+	memset(ps->zero_area, 0, len);
+
+	return 0;
+}
+
+static void free_area(struct pstore *ps)
+{
+	vfree(ps->area);
+	ps->area = NULL;
+	vfree(ps->zero_area);
+	ps->zero_area = NULL;
+}
+
+struct mdata_req {
+	struct dm_io_region *where;
+	struct dm_io_request *io_req;
+	struct work_struct work;
+	int result;
+};
+
+static void do_metadata(struct work_struct *work)
+{
+	struct mdata_req *req = container_of(work, struct mdata_req, work);
+
+	req->result = dm_io(req->io_req, 1, req->where, NULL);
+}
+
+/*
+ * Read or write a chunk aligned and sized block of data from a device.
+ */
+static int chunk_io(struct pstore *ps, chunk_t chunk, int rw, int metadata)
+{
+	struct dm_io_region where = {
+		.bdev = ps->snap->cow->bdev,
+		.sector = ps->snap->chunk_size * chunk,
+		.count = ps->snap->chunk_size,
+	};
+	struct dm_io_request io_req = {
+		.bi_rw = rw,
+		.mem.type = DM_IO_VMA,
+		.mem.ptr.vma = ps->area,
+		.client = ps->io_client,
+		.notify.fn = NULL,
+	};
+	struct mdata_req req;
+
+	if (!metadata)
+		return dm_io(&io_req, 1, &where, NULL);
+
+	req.where = &where;
+	req.io_req = &io_req;
+
+	/*
+	 * Issue the synchronous I/O from a different thread
+	 * to avoid generic_make_request recursion.
+	 */
+	INIT_WORK(&req.work, do_metadata);
+	queue_work(ps->metadata_wq, &req.work);
+	flush_workqueue(ps->metadata_wq);
+
+	return req.result;
+}
+
+/*
+ * Convert a metadata area index to a chunk index.
+ */
+static chunk_t area_location(struct pstore *ps, chunk_t area)
+{
+	return 1 + ((ps->exceptions_per_area + 1) * area);
+}
+
+/*
+ * Read or write a metadata area.  Remembering to skip the first
+ * chunk which holds the header.
+ */
+static int area_io(struct pstore *ps, int rw)
+{
+	int r;
+	chunk_t chunk;
+
+	chunk = area_location(ps, ps->current_area);
+
+	r = chunk_io(ps, chunk, rw, 0);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void zero_memory_area(struct pstore *ps)
+{
+	memset(ps->area, 0, ps->snap->chunk_size << SECTOR_SHIFT);
+}
+
+static int zero_disk_area(struct pstore *ps, chunk_t area)
+{
+	struct dm_io_region where = {
+		.bdev = ps->snap->cow->bdev,
+		.sector = ps->snap->chunk_size * area_location(ps, area),
+		.count = ps->snap->chunk_size,
+	};
+	struct dm_io_request io_req = {
+		.bi_rw = WRITE,
+		.mem.type = DM_IO_VMA,
+		.mem.ptr.vma = ps->zero_area,
+		.client = ps->io_client,
+		.notify.fn = NULL,
+	};
+
+	return dm_io(&io_req, 1, &where, NULL);
+}
+
+static int read_header(struct pstore *ps, int *new_snapshot)
+{
+	int r;
+	struct disk_header *dh;
+	chunk_t chunk_size;
+	int chunk_size_supplied = 1;
+
+	/*
+	 * Use default chunk size (or hardsect_size, if larger) if none supplied
+	 */
+	if (!ps->snap->chunk_size) {
+		ps->snap->chunk_size = max(DM_CHUNK_SIZE_DEFAULT_SECTORS,
+		    bdev_hardsect_size(ps->snap->cow->bdev) >> 9);
+		ps->snap->chunk_mask = ps->snap->chunk_size - 1;
+		ps->snap->chunk_shift = ffs(ps->snap->chunk_size) - 1;
+		chunk_size_supplied = 0;
+	}
+
+	ps->io_client = dm_io_client_create(sectors_to_pages(ps->snap->
+							     chunk_size));
+	if (IS_ERR(ps->io_client))
+		return PTR_ERR(ps->io_client);
+
+	r = alloc_area(ps);
+	if (r)
+		return r;
+
+	r = chunk_io(ps, 0, READ, 1);
+	if (r)
+		goto bad;
+
+	dh = (struct disk_header *) ps->area;
+
+	if (le32_to_cpu(dh->magic) == 0) {
+		*new_snapshot = 1;
+		return 0;
+	}
+
+	if (le32_to_cpu(dh->magic) != SNAP_MAGIC) {
+		DMWARN("Invalid or corrupt snapshot");
+		r = -ENXIO;
+		goto bad;
+	}
+
+	*new_snapshot = 0;
+	ps->valid = le32_to_cpu(dh->valid);
+	ps->version = le32_to_cpu(dh->version);
+	chunk_size = le32_to_cpu(dh->chunk_size);
+
+	if (!chunk_size_supplied || ps->snap->chunk_size == chunk_size)
+		return 0;
+
+	DMWARN("chunk size %llu in device metadata overrides "
+	       "table chunk size of %llu.",
+	       (unsigned long long)chunk_size,
+	       (unsigned long long)ps->snap->chunk_size);
+
+	/* We had a bogus chunk_size. Fix stuff up. */
+	free_area(ps);
+
+	ps->snap->chunk_size = chunk_size;
+	ps->snap->chunk_mask = chunk_size - 1;
+	ps->snap->chunk_shift = ffs(chunk_size) - 1;
+
+	r = dm_io_client_resize(sectors_to_pages(ps->snap->chunk_size),
+				ps->io_client);
+	if (r)
+		return r;
+
+	r = alloc_area(ps);
+	return r;
+
+bad:
+	free_area(ps);
+	return r;
+}
+
+static int write_header(struct pstore *ps)
+{
+	struct disk_header *dh;
+
+	memset(ps->area, 0, ps->snap->chunk_size << SECTOR_SHIFT);
+
+	dh = (struct disk_header *) ps->area;
+	dh->magic = cpu_to_le32(SNAP_MAGIC);
+	dh->valid = cpu_to_le32(ps->valid);
+	dh->version = cpu_to_le32(ps->version);
+	dh->chunk_size = cpu_to_le32(ps->snap->chunk_size);
+
+	return chunk_io(ps, 0, WRITE, 1);
+}
+
+/*
+ * Access functions for the disk exceptions, these do the endian conversions.
+ */
+static struct disk_exception *get_exception(struct pstore *ps, uint32_t index)
+{
+	BUG_ON(index >= ps->exceptions_per_area);
+
+	return ((struct disk_exception *) ps->area) + index;
+}
+
+static void read_exception(struct pstore *ps,
+			   uint32_t index, struct disk_exception *result)
+{
+	struct disk_exception *e = get_exception(ps, index);
+
+	/* copy it */
+	result->old_chunk = le64_to_cpu(e->old_chunk);
+	result->new_chunk = le64_to_cpu(e->new_chunk);
+}
+
+static void write_exception(struct pstore *ps,
+			    uint32_t index, struct disk_exception *de)
+{
+	struct disk_exception *e = get_exception(ps, index);
+
+	/* copy it */
+	e->old_chunk = cpu_to_le64(de->old_chunk);
+	e->new_chunk = cpu_to_le64(de->new_chunk);
+}
+
+/*
+ * Registers the exceptions that are present in the current area.
+ * 'full' is filled in to indicate if the area has been
+ * filled.
+ */
+static int insert_exceptions(struct pstore *ps,
+			     int (*callback)(void *callback_context,
+					     chunk_t old, chunk_t new),
+			     void *callback_context,
+			     int *full)
+{
+	int r;
+	unsigned int i;
+	struct disk_exception de;
+
+	/* presume the area is full */
+	*full = 1;
+
+	for (i = 0; i < ps->exceptions_per_area; i++) {
+		read_exception(ps, i, &de);
+
+		/*
+		 * If the new_chunk is pointing at the start of
+		 * the COW device, where the first metadata area
+		 * is we know that we've hit the end of the
+		 * exceptions.  Therefore the area is not full.
+		 */
+		if (de.new_chunk == 0LL) {
+			ps->current_committed = i;
+			*full = 0;
+			break;
+		}
+
+		/*
+		 * Keep track of the start of the free chunks.
+		 */
+		if (ps->next_free <= de.new_chunk)
+			ps->next_free = de.new_chunk + 1;
+
+		/*
+		 * Otherwise we add the exception to the snapshot.
+		 */
+		r = callback(callback_context, de.old_chunk, de.new_chunk);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int read_exceptions(struct pstore *ps,
+			   int (*callback)(void *callback_context, chunk_t old,
+					   chunk_t new),
+			   void *callback_context)
+{
+	int r, full = 1;
+
+	/*
+	 * Keeping reading chunks and inserting exceptions until
+	 * we find a partially full area.
+	 */
+	for (ps->current_area = 0; full; ps->current_area++) {
+		r = area_io(ps, READ);
+		if (r)
+			return r;
+
+		r = insert_exceptions(ps, callback, callback_context, &full);
+		if (r)
+			return r;
+	}
+
+	ps->current_area--;
+
+	return 0;
+}
+
+static struct pstore *get_info(struct dm_exception_store *store)
+{
+	return (struct pstore *) store->context;
+}
+
+static void persistent_fraction_full(struct dm_exception_store *store,
+				     sector_t *numerator, sector_t *denominator)
+{
+	*numerator = get_info(store)->next_free * store->snap->chunk_size;
+	*denominator = get_dev_size(store->snap->cow->bdev);
+}
+
+static void persistent_destroy(struct dm_exception_store *store)
+{
+	struct pstore *ps = get_info(store);
+
+	destroy_workqueue(ps->metadata_wq);
+	dm_io_client_destroy(ps->io_client);
+	vfree(ps->callbacks);
+	free_area(ps);
+	kfree(ps);
+}
+
+static int persistent_read_metadata(struct dm_exception_store *store,
+				    int (*callback)(void *callback_context,
+						    chunk_t old, chunk_t new),
+				    void *callback_context)
+{
+	int r, uninitialized_var(new_snapshot);
+	struct pstore *ps = get_info(store);
+
+	/*
+	 * Read the snapshot header.
+	 */
+	r = read_header(ps, &new_snapshot);
+	if (r)
+		return r;
+
+	/*
+	 * Now we know correct chunk_size, complete the initialisation.
+	 */
+	ps->exceptions_per_area = (ps->snap->chunk_size << SECTOR_SHIFT) /
+				  sizeof(struct disk_exception);
+	ps->callbacks = dm_vcalloc(ps->exceptions_per_area,
+			sizeof(*ps->callbacks));
+	if (!ps->callbacks)
+		return -ENOMEM;
+
+	/*
+	 * Do we need to setup a new snapshot ?
+	 */
+	if (new_snapshot) {
+		r = write_header(ps);
+		if (r) {
+			DMWARN("write_header failed");
+			return r;
+		}
+
+		ps->current_area = 0;
+		zero_memory_area(ps);
+		r = zero_disk_area(ps, 0);
+		if (r) {
+			DMWARN("zero_disk_area(0) failed");
+			return r;
+		}
+	} else {
+		/*
+		 * Sanity checks.
+		 */
+		if (ps->version != SNAPSHOT_DISK_VERSION) {
+			DMWARN("unable to handle snapshot disk version %d",
+			       ps->version);
+			return -EINVAL;
+		}
+
+		/*
+		 * Metadata are valid, but snapshot is invalidated
+		 */
+		if (!ps->valid)
+			return 1;
+
+		/*
+		 * Read the metadata.
+		 */
+		r = read_exceptions(ps, callback, callback_context);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int persistent_prepare_exception(struct dm_exception_store *store,
+					struct dm_snap_exception *e)
+{
+	struct pstore *ps = get_info(store);
+	uint32_t stride;
+	chunk_t next_free;
+	sector_t size = get_dev_size(store->snap->cow->bdev);
+
+	/* Is there enough room ? */
+	if (size < ((ps->next_free + 1) * store->snap->chunk_size))
+		return -ENOSPC;
+
+	e->new_chunk = ps->next_free;
+
+	/*
+	 * Move onto the next free pending, making sure to take
+	 * into account the location of the metadata chunks.
+	 */
+	stride = (ps->exceptions_per_area + 1);
+	next_free = ++ps->next_free;
+	if (sector_div(next_free, stride) == 1)
+		ps->next_free++;
+
+	atomic_inc(&ps->pending_count);
+	return 0;
+}
+
+static void persistent_commit_exception(struct dm_exception_store *store,
+					struct dm_snap_exception *e,
+					void (*callback) (void *, int success),
+					void *callback_context)
+{
+	unsigned int i;
+	struct pstore *ps = get_info(store);
+	struct disk_exception de;
+	struct commit_callback *cb;
+
+	de.old_chunk = e->old_chunk;
+	de.new_chunk = e->new_chunk;
+	write_exception(ps, ps->current_committed++, &de);
+
+	/*
+	 * Add the callback to the back of the array.  This code
+	 * is the only place where the callback array is
+	 * manipulated, and we know that it will never be called
+	 * multiple times concurrently.
+	 */
+	cb = ps->callbacks + ps->callback_count++;
+	cb->callback = callback;
+	cb->context = callback_context;
+
+	/*
+	 * If there are exceptions in flight and we have not yet
+	 * filled this metadata area there's nothing more to do.
+	 */
+	if (!atomic_dec_and_test(&ps->pending_count) &&
+	    (ps->current_committed != ps->exceptions_per_area))
+		return;
+
+	/*
+	 * If we completely filled the current area, then wipe the next one.
+	 */
+	if ((ps->current_committed == ps->exceptions_per_area) &&
+	     zero_disk_area(ps, ps->current_area + 1))
+		ps->valid = 0;
+
+	/*
+	 * Commit exceptions to disk.
+	 */
+	if (ps->valid && area_io(ps, WRITE))
+		ps->valid = 0;
+
+	/*
+	 * Advance to the next area if this one is full.
+	 */
+	if (ps->current_committed == ps->exceptions_per_area) {
+		ps->current_committed = 0;
+		ps->current_area++;
+		zero_memory_area(ps);
+	}
+
+	for (i = 0; i < ps->callback_count; i++) {
+		cb = ps->callbacks + i;
+		cb->callback(cb->context, ps->valid);
+	}
+
+	ps->callback_count = 0;
+}
+
+static void persistent_drop_snapshot(struct dm_exception_store *store)
+{
+	struct pstore *ps = get_info(store);
+
+	ps->valid = 0;
+	if (write_header(ps))
+		DMWARN("write header failed");
+}
+
+int dm_create_persistent(struct dm_exception_store *store)
+{
+	struct pstore *ps;
+
+	/* allocate the pstore */
+	ps = kmalloc(sizeof(*ps), GFP_KERNEL);
+	if (!ps)
+		return -ENOMEM;
+
+	ps->snap = store->snap;
+	ps->valid = 1;
+	ps->version = SNAPSHOT_DISK_VERSION;
+	ps->area = NULL;
+	ps->next_free = 2;	/* skipping the header and first area */
+	ps->current_committed = 0;
+
+	ps->callback_count = 0;
+	atomic_set(&ps->pending_count, 0);
+	ps->callbacks = NULL;
+
+	ps->metadata_wq = create_singlethread_workqueue("ksnaphd");
+	if (!ps->metadata_wq) {
+		kfree(ps);
+		DMERR("couldn't start header metadata update thread");
+		return -ENOMEM;
+	}
+
+	store->destroy = persistent_destroy;
+	store->read_metadata = persistent_read_metadata;
+	store->prepare_exception = persistent_prepare_exception;
+	store->commit_exception = persistent_commit_exception;
+	store->drop_snapshot = persistent_drop_snapshot;
+	store->fraction_full = persistent_fraction_full;
+	store->context = ps;
+
+	return 0;
+}
+
+int dm_persistent_snapshot_init(void)
+{
+	return 0;
+}
+
+void dm_persistent_snapshot_exit(void)
+{
+}
diff --git a/drivers/md/dm-snap-transient.c b/drivers/md/dm-snap-transient.c
new file mode 100644
index 00000000000..7f6e2e6dcb0
--- /dev/null
+++ b/drivers/md/dm-snap-transient.c
@@ -0,0 +1,98 @@
+/*
+ * Copyright (C) 2001-2002 Sistina Software (UK) Limited.
+ * Copyright (C) 2006-2008 Red Hat GmbH
+ *
+ * This file is released under the GPL.
+ */
+
+#include "dm-exception-store.h"
+#include "dm-snap.h"
+
+#include <linux/mm.h>
+#include <linux/pagemap.h>
+#include <linux/vmalloc.h>
+#include <linux/slab.h>
+#include <linux/dm-io.h>
+
+#define DM_MSG_PREFIX "transient snapshot"
+
+/*-----------------------------------------------------------------
+ * Implementation of the store for non-persistent snapshots.
+ *---------------------------------------------------------------*/
+struct transient_c {
+	sector_t next_free;
+};
+
+static void transient_destroy(struct dm_exception_store *store)
+{
+	kfree(store->context);
+}
+
+static int transient_read_metadata(struct dm_exception_store *store,
+				   int (*callback)(void *callback_context,
+						   chunk_t old, chunk_t new),
+				   void *callback_context)
+{
+	return 0;
+}
+
+static int transient_prepare_exception(struct dm_exception_store *store,
+				       struct dm_snap_exception *e)
+{
+	struct transient_c *tc = (struct transient_c *) store->context;
+	sector_t size = get_dev_size(store->snap->cow->bdev);
+
+	if (size < (tc->next_free + store->snap->chunk_size))
+		return -1;
+
+	e->new_chunk = sector_to_chunk(store->snap, tc->next_free);
+	tc->next_free += store->snap->chunk_size;
+
+	return 0;
+}
+
+static void transient_commit_exception(struct dm_exception_store *store,
+				       struct dm_snap_exception *e,
+				       void (*callback) (void *, int success),
+				       void *callback_context)
+{
+	/* Just succeed */
+	callback(callback_context, 1);
+}
+
+static void transient_fraction_full(struct dm_exception_store *store,
+				    sector_t *numerator, sector_t *denominator)
+{
+	*numerator = ((struct transient_c *) store->context)->next_free;
+	*denominator = get_dev_size(store->snap->cow->bdev);
+}
+
+int dm_create_transient(struct dm_exception_store *store)
+{
+	struct transient_c *tc;
+
+	store->destroy = transient_destroy;
+	store->read_metadata = transient_read_metadata;
+	store->prepare_exception = transient_prepare_exception;
+	store->commit_exception = transient_commit_exception;
+	store->drop_snapshot = NULL;
+	store->fraction_full = transient_fraction_full;
+
+	tc = kmalloc(sizeof(struct transient_c), GFP_KERNEL);
+	if (!tc)
+		return -ENOMEM;
+
+	tc->next_free = 0;
+	store->context = tc;
+
+	return 0;
+}
+
+int dm_transient_snapshot_init(void)
+{
+	return 0;
+}
+
+void dm_transient_snapshot_exit(void)
+{
+}
diff --git a/drivers/md/dm-snap.c b/drivers/md/dm-snap.c
index 6c96db26b87..65ff82ff124 100644
--- a/drivers/md/dm-snap.c
+++ b/drivers/md/dm-snap.c
@@ -9,6 +9,7 @@
 #include <linux/blkdev.h>
 #include <linux/ctype.h>
 #include <linux/device-mapper.h>
+#include <linux/delay.h>
 #include <linux/fs.h>
 #include <linux/init.h>
 #include <linux/kdev_t.h>
@@ -20,6 +21,7 @@
 #include <linux/log2.h>
 #include <linux/dm-kcopyd.h>
 
+#include "dm-exception-store.h"
 #include "dm-snap.h"
 #include "dm-bio-list.h"
 
@@ -428,8 +430,13 @@ out:
 	list_add(&new_e->hash_list, e ? &e->hash_list : l);
 }
 
-int dm_add_exception(struct dm_snapshot *s, chunk_t old, chunk_t new)
+/*
+ * Callback used by the exception stores to load exceptions when
+ * initialising.
+ */
+static int dm_add_exception(void *context, chunk_t old, chunk_t new)
 {
+	struct dm_snapshot *s = context;
 	struct dm_snap_exception *e;
 
 	e = alloc_exception();
@@ -658,7 +665,7 @@ static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 	spin_lock_init(&s->tracked_chunk_lock);
 
 	/* Metadata must only be loaded into one table at once */
-	r = s->store.read_metadata(&s->store);
+	r = s->store.read_metadata(&s->store, dm_add_exception, (void *)s);
 	if (r < 0) {
 		ti->error = "Failed to read snapshot metadata";
 		goto bad_load_and_register;
@@ -735,7 +742,7 @@ static void snapshot_dtr(struct dm_target *ti)
 	unregister_snapshot(s);
 
 	while (atomic_read(&s->pending_exceptions_count))
-		yield();
+		msleep(1);
 	/*
 	 * Ensure instructions in mempool_destroy aren't reordered
 	 * before atomic_read.
@@ -888,10 +895,10 @@ static void pending_complete(struct dm_snap_pending_exception *pe, int success)
 
 	/*
 	 * Check for conflicting reads. This is extremely improbable,
-	 * so yield() is sufficient and there is no need for a wait queue.
+	 * so msleep(1) is sufficient and there is no need for a wait queue.
 	 */
 	while (__chunk_is_tracked(s, pe->e.old_chunk))
-		yield();
+		msleep(1);
 
 	/*
 	 * Add a proper exception, and remove the
@@ -1404,6 +1411,12 @@ static int __init dm_snapshot_init(void)
 {
 	int r;
 
+	r = dm_exception_store_init();
+	if (r) {
+		DMERR("Failed to initialize exception stores");
+		return r;
+	}
+
 	r = dm_register_target(&snapshot_target);
 	if (r) {
 		DMERR("snapshot target register failed %d", r);
@@ -1452,39 +1465,34 @@ static int __init dm_snapshot_init(void)
 
 	return 0;
 
-      bad_pending_pool:
+bad_pending_pool:
 	kmem_cache_destroy(tracked_chunk_cache);
-      bad5:
+bad5:
 	kmem_cache_destroy(pending_cache);
-      bad4:
+bad4:
 	kmem_cache_destroy(exception_cache);
-      bad3:
+bad3:
 	exit_origin_hash();
-      bad2:
+bad2:
 	dm_unregister_target(&origin_target);
-      bad1:
+bad1:
 	dm_unregister_target(&snapshot_target);
 	return r;
 }
 
 static void __exit dm_snapshot_exit(void)
 {
-	int r;
-
 	destroy_workqueue(ksnapd);
 
-	r = dm_unregister_target(&snapshot_target);
-	if (r)
-		DMERR("snapshot unregister failed %d", r);
-
-	r = dm_unregister_target(&origin_target);
-	if (r)
-		DMERR("origin unregister failed %d", r);
+	dm_unregister_target(&snapshot_target);
+	dm_unregister_target(&origin_target);
 
 	exit_origin_hash();
 	kmem_cache_destroy(pending_cache);
 	kmem_cache_destroy(exception_cache);
 	kmem_cache_destroy(tracked_chunk_cache);
+
+	dm_exception_store_exit();
 }
 
 /* Module hooks */
diff --git a/drivers/md/dm-snap.h b/drivers/md/dm-snap.h
index 99c0106ede2..d9e62b43cf8 100644
--- a/drivers/md/dm-snap.h
+++ b/drivers/md/dm-snap.h
@@ -1,6 +1,4 @@
 /*
- * dm-snapshot.c
- *
  * Copyright (C) 2001-2002 Sistina Software (UK) Limited.
  *
  * This file is released under the GPL.
@@ -10,6 +8,7 @@
 #define DM_SNAPSHOT_H
 
 #include <linux/device-mapper.h>
+#include "dm-exception-store.h"
 #include "dm-bio-list.h"
 #include <linux/blkdev.h>
 #include <linux/workqueue.h>
@@ -20,116 +19,6 @@ struct exception_table {
 	struct list_head *table;
 };
 
-/*
- * The snapshot code deals with largish chunks of the disk at a
- * time. Typically 32k - 512k.
- */
-typedef sector_t chunk_t;
-
-/*
- * An exception is used where an old chunk of data has been
- * replaced by a new one.
- * If chunk_t is 64 bits in size, the top 8 bits of new_chunk hold the number
- * of chunks that follow contiguously.  Remaining bits hold the number of the
- * chunk within the device.
- */
-struct dm_snap_exception {
-	struct list_head hash_list;
-
-	chunk_t old_chunk;
-	chunk_t new_chunk;
-};
-
-/*
- * Funtions to manipulate consecutive chunks
- */
-#  if defined(CONFIG_LBD) || (BITS_PER_LONG == 64)
-#    define DM_CHUNK_CONSECUTIVE_BITS 8
-#    define DM_CHUNK_NUMBER_BITS 56
-
-static inline chunk_t dm_chunk_number(chunk_t chunk)
-{
-	return chunk & (chunk_t)((1ULL << DM_CHUNK_NUMBER_BITS) - 1ULL);
-}
-
-static inline unsigned dm_consecutive_chunk_count(struct dm_snap_exception *e)
-{
-	return e->new_chunk >> DM_CHUNK_NUMBER_BITS;
-}
-
-static inline void dm_consecutive_chunk_count_inc(struct dm_snap_exception *e)
-{
-	e->new_chunk += (1ULL << DM_CHUNK_NUMBER_BITS);
-
-	BUG_ON(!dm_consecutive_chunk_count(e));
-}
-
-#  else
-#    define DM_CHUNK_CONSECUTIVE_BITS 0
-
-static inline chunk_t dm_chunk_number(chunk_t chunk)
-{
-	return chunk;
-}
-
-static inline unsigned dm_consecutive_chunk_count(struct dm_snap_exception *e)
-{
-	return 0;
-}
-
-static inline void dm_consecutive_chunk_count_inc(struct dm_snap_exception *e)
-{
-}
-
-#  endif
-
-/*
- * Abstraction to handle the meta/layout of exception stores (the
- * COW device).
- */
-struct exception_store {
-
-	/*
-	 * Destroys this object when you've finished with it.
-	 */
-	void (*destroy) (struct exception_store *store);
-
-	/*
-	 * The target shouldn't read the COW device until this is
-	 * called.
-	 */
-	int (*read_metadata) (struct exception_store *store);
-
-	/*
-	 * Find somewhere to store the next exception.
-	 */
-	int (*prepare_exception) (struct exception_store *store,
-				  struct dm_snap_exception *e);
-
-	/*
-	 * Update the metadata with this exception.
-	 */
-	void (*commit_exception) (struct exception_store *store,
-				  struct dm_snap_exception *e,
-				  void (*callback) (void *, int success),
-				  void *callback_context);
-
-	/*
-	 * The snapshot is invalid, note this in the metadata.
-	 */
-	void (*drop_snapshot) (struct exception_store *store);
-
-	/*
-	 * Return how full the snapshot is.
-	 */
-	void (*fraction_full) (struct exception_store *store,
-			       sector_t *numerator,
-			       sector_t *denominator);
-
-	struct dm_snapshot *snap;
-	void *context;
-};
-
 #define DM_TRACKED_CHUNK_HASH_SIZE	16
 #define DM_TRACKED_CHUNK_HASH(x)	((unsigned long)(x) & \
 					 (DM_TRACKED_CHUNK_HASH_SIZE - 1))
@@ -172,7 +61,7 @@ struct dm_snapshot {
 	spinlock_t pe_lock;
 
 	/* The on disk metadata handler */
-	struct exception_store store;
+	struct dm_exception_store store;
 
 	struct dm_kcopyd_client *kcopyd_client;
 
@@ -187,20 +76,6 @@ struct dm_snapshot {
 };
 
 /*
- * Used by the exception stores to load exceptions hen
- * initialising.
- */
-int dm_add_exception(struct dm_snapshot *s, chunk_t old, chunk_t new);
-
-/*
- * Constructor and destructor for the default persistent
- * store.
- */
-int dm_create_persistent(struct exception_store *store);
-
-int dm_create_transient(struct exception_store *store);
-
-/*
  * Return the number of sectors in the device.
  */
 static inline sector_t get_dev_size(struct block_device *bdev)
diff --git a/drivers/md/dm-stripe.c b/drivers/md/dm-stripe.c
index a2d068dbe9e..41569bc60ab 100644
--- a/drivers/md/dm-stripe.c
+++ b/drivers/md/dm-stripe.c
@@ -320,8 +320,10 @@ int __init dm_stripe_init(void)
 	int r;
 
 	r = dm_register_target(&stripe_target);
-	if (r < 0)
+	if (r < 0) {
 		DMWARN("target registration failed");
+		return r;
+	}
 
 	kstriped = create_singlethread_workqueue("kstriped");
 	if (!kstriped) {
@@ -335,9 +337,7 @@ int __init dm_stripe_init(void)
 
 void dm_stripe_exit(void)
 {
-	if (dm_unregister_target(&stripe_target))
-		DMWARN("target unregistration failed");
-
+	dm_unregister_target(&stripe_target);
 	destroy_workqueue(kstriped);
 
 	return;
diff --git a/drivers/md/dm-sysfs.c b/drivers/md/dm-sysfs.c
new file mode 100644
index 00000000000..a2a45e6c7c8
--- /dev/null
+++ b/drivers/md/dm-sysfs.c
@@ -0,0 +1,99 @@
+/*
+ * Copyright (C) 2008 Red Hat, Inc. All rights reserved.
+ *
+ * This file is released under the GPL.
+ */
+
+#include <linux/sysfs.h>
+#include <linux/dm-ioctl.h>
+#include "dm.h"
+
+struct dm_sysfs_attr {
+	struct attribute attr;
+	ssize_t (*show)(struct mapped_device *, char *);
+	ssize_t (*store)(struct mapped_device *, char *);
+};
+
+#define DM_ATTR_RO(_name) \
+struct dm_sysfs_attr dm_attr_##_name = \
+	__ATTR(_name, S_IRUGO, dm_attr_##_name##_show, NULL)
+
+static ssize_t dm_attr_show(struct kobject *kobj, struct attribute *attr,
+			    char *page)
+{
+	struct dm_sysfs_attr *dm_attr;
+	struct mapped_device *md;
+	ssize_t ret;
+
+	dm_attr = container_of(attr, struct dm_sysfs_attr, attr);
+	if (!dm_attr->show)
+		return -EIO;
+
+	md = dm_get_from_kobject(kobj);
+	if (!md)
+		return -EINVAL;
+
+	ret = dm_attr->show(md, page);
+	dm_put(md);
+
+	return ret;
+}
+
+static ssize_t dm_attr_name_show(struct mapped_device *md, char *buf)
+{
+	if (dm_copy_name_and_uuid(md, buf, NULL))
+		return -EIO;
+
+	strcat(buf, "\n");
+	return strlen(buf);
+}
+
+static ssize_t dm_attr_uuid_show(struct mapped_device *md, char *buf)
+{
+	if (dm_copy_name_and_uuid(md, NULL, buf))
+		return -EIO;
+
+	strcat(buf, "\n");
+	return strlen(buf);
+}
+
+static DM_ATTR_RO(name);
+static DM_ATTR_RO(uuid);
+
+static struct attribute *dm_attrs[] = {
+	&dm_attr_name.attr,
+	&dm_attr_uuid.attr,
+	NULL,
+};
+
+static struct sysfs_ops dm_sysfs_ops = {
+	.show	= dm_attr_show,
+};
+
+/*
+ * dm kobject is embedded in mapped_device structure
+ * no need to define release function here
+ */
+static struct kobj_type dm_ktype = {
+	.sysfs_ops	= &dm_sysfs_ops,
+	.default_attrs	= dm_attrs,
+};
+
+/*
+ * Initialize kobj
+ * because nobody using md yet, no need to call explicit dm_get/put
+ */
+int dm_sysfs_init(struct mapped_device *md)
+{
+	return kobject_init_and_add(dm_kobject(md), &dm_ktype,
+				    &disk_to_dev(dm_disk(md))->kobj,
+				    "%s", "dm");
+}
+
+/*
+ * Remove kobj, called after all references removed
+ */
+void dm_sysfs_exit(struct mapped_device *md)
+{
+	kobject_put(dm_kobject(md));
+}
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index a63161aec48..2fd66c30f7f 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2001 Sistina Software (UK) Limited.
- * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
  *
  * This file is released under the GPL.
  */
@@ -15,6 +15,7 @@
 #include <linux/slab.h>
 #include <linux/interrupt.h>
 #include <linux/mutex.h>
+#include <linux/delay.h>
 #include <asm/atomic.h>
 
 #define DM_MSG_PREFIX "table"
@@ -24,6 +25,19 @@
 #define KEYS_PER_NODE (NODE_SIZE / sizeof(sector_t))
 #define CHILDREN_PER_NODE (KEYS_PER_NODE + 1)
 
+/*
+ * The table has always exactly one reference from either mapped_device->map
+ * or hash_cell->new_map. This reference is not counted in table->holders.
+ * A pair of dm_create_table/dm_destroy_table functions is used for table
+ * creation/destruction.
+ *
+ * Temporary references from the other code increase table->holders. A pair
+ * of dm_table_get/dm_table_put functions is used to manipulate it.
+ *
+ * When the table is about to be destroyed, we wait for table->holders to
+ * drop to zero.
+ */
+
 struct dm_table {
 	struct mapped_device *md;
 	atomic_t holders;
@@ -38,6 +52,8 @@ struct dm_table {
 	sector_t *highs;
 	struct dm_target *targets;
 
+	unsigned barriers_supported:1;
+
 	/*
 	 * Indicates the rw permissions for the new logical
 	 * device.  This should be a combination of FMODE_READ
@@ -226,7 +242,8 @@ int dm_table_create(struct dm_table **result, fmode_t mode,
 		return -ENOMEM;
 
 	INIT_LIST_HEAD(&t->devices);
-	atomic_set(&t->holders, 1);
+	atomic_set(&t->holders, 0);
+	t->barriers_supported = 1;
 
 	if (!num_targets)
 		num_targets = KEYS_PER_NODE;
@@ -256,10 +273,14 @@ static void free_devices(struct list_head *devices)
 	}
 }
 
-static void table_destroy(struct dm_table *t)
+void dm_table_destroy(struct dm_table *t)
 {
 	unsigned int i;
 
+	while (atomic_read(&t->holders))
+		msleep(1);
+	smp_mb();
+
 	/* free the indexes (see dm_table_complete) */
 	if (t->depth >= 2)
 		vfree(t->index[t->depth - 2]);
@@ -297,8 +318,8 @@ void dm_table_put(struct dm_table *t)
 	if (!t)
 		return;
 
-	if (atomic_dec_and_test(&t->holders))
-		table_destroy(t);
+	smp_mb__before_atomic_dec();
+	atomic_dec(&t->holders);
 }
 
 /*
@@ -668,7 +689,7 @@ static void check_for_valid_limits(struct io_restrictions *rs)
 	if (!rs->max_segment_size)
 		rs->max_segment_size = MAX_SEGMENT_SIZE;
 	if (!rs->seg_boundary_mask)
-		rs->seg_boundary_mask = -1;
+		rs->seg_boundary_mask = BLK_SEG_BOUNDARY_MASK;
 	if (!rs->bounce_pfn)
 		rs->bounce_pfn = -1;
 }
@@ -728,6 +749,10 @@ int dm_table_add_target(struct dm_table *t, const char *type,
 	/* FIXME: the plan is to combine high here and then have
 	 * the merge fn apply the target level restrictions. */
 	combine_restrictions_low(&t->limits, &tgt->limits);
+
+	if (!(tgt->type->features & DM_TARGET_SUPPORTS_BARRIERS))
+		t->barriers_supported = 0;
+
 	return 0;
 
  bad:
@@ -772,6 +797,12 @@ int dm_table_complete(struct dm_table *t)
 
 	check_for_valid_limits(&t->limits);
 
+	/*
+	 * We only support barriers if there is exactly one underlying device.
+	 */
+	if (!list_is_singular(&t->devices))
+		t->barriers_supported = 0;
+
 	/* how many indexes will the btree have ? */
 	leaf_nodes = dm_div_up(t->num_targets, KEYS_PER_NODE);
 	t->depth = 1 + int_log(leaf_nodes, CHILDREN_PER_NODE);
@@ -986,6 +1017,12 @@ struct mapped_device *dm_table_get_md(struct dm_table *t)
 	return t->md;
 }
 
+int dm_table_barrier_ok(struct dm_table *t)
+{
+	return t->barriers_supported;
+}
+EXPORT_SYMBOL(dm_table_barrier_ok);
+
 EXPORT_SYMBOL(dm_vcalloc);
 EXPORT_SYMBOL(dm_get_device);
 EXPORT_SYMBOL(dm_put_device);
diff --git a/drivers/md/dm-target.c b/drivers/md/dm-target.c
index 835cf95b857..7decf10006e 100644
--- a/drivers/md/dm-target.c
+++ b/drivers/md/dm-target.c
@@ -130,26 +130,26 @@ int dm_register_target(struct target_type *t)
 	return rv;
 }
 
-int dm_unregister_target(struct target_type *t)
+void dm_unregister_target(struct target_type *t)
 {
 	struct tt_internal *ti;
 
 	down_write(&_lock);
 	if (!(ti = __find_target_type(t->name))) {
-		up_write(&_lock);
-		return -EINVAL;
+		DMCRIT("Unregistering unrecognised target: %s", t->name);
+		BUG();
 	}
 
 	if (ti->use) {
-		up_write(&_lock);
-		return -ETXTBSY;
+		DMCRIT("Attempt to unregister target still in use: %s",
+		       t->name);
+		BUG();
 	}
 
 	list_del(&ti->list);
 	kfree(ti);
 
 	up_write(&_lock);
-	return 0;
 }
 
 /*
@@ -187,8 +187,7 @@ int __init dm_target_init(void)
 
 void dm_target_exit(void)
 {
-	if (dm_unregister_target(&error_target))
-		DMWARN("error target unregistration failed");
+	dm_unregister_target(&error_target);
 }
 
 EXPORT_SYMBOL(dm_register_target);
diff --git a/drivers/md/dm-zero.c b/drivers/md/dm-zero.c
index cdbf126ec10..bbc97030c0c 100644
--- a/drivers/md/dm-zero.c
+++ b/drivers/md/dm-zero.c
@@ -69,10 +69,7 @@ static int __init dm_zero_init(void)
 
 static void __exit dm_zero_exit(void)
 {
-	int r = dm_unregister_target(&zero_target);
-
-	if (r < 0)
-		DMERR("unregister failed %d", r);
+	dm_unregister_target(&zero_target);
 }
 
 module_init(dm_zero_init)
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 6963ad14840..51ba1db4b3e 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2001, 2002 Sistina Software (UK) Limited.
- * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
  *
  * This file is released under the GPL.
  */
@@ -21,6 +21,7 @@
 #include <linux/idr.h>
 #include <linux/hdreg.h>
 #include <linux/blktrace_api.h>
+#include <trace/block.h>
 
 #define DM_MSG_PREFIX "core"
 
@@ -31,6 +32,7 @@ static unsigned int _major = 0;
 
 static DEFINE_SPINLOCK(_minor_lock);
 /*
+ * For bio-based dm.
  * One of these is allocated per bio.
  */
 struct dm_io {
@@ -42,6 +44,7 @@ struct dm_io {
 };
 
 /*
+ * For bio-based dm.
  * One of these is allocated per target within a bio.  Hopefully
  * this will be simplified out one day.
  */
@@ -51,6 +54,29 @@ struct dm_target_io {
 	union map_info info;
 };
 
+DEFINE_TRACE(block_bio_complete);
+
+/*
+ * For request-based dm.
+ * One of these is allocated per request.
+ */
+struct dm_rq_target_io {
+	struct mapped_device *md;
+	struct dm_target *ti;
+	struct request *orig, clone;
+	int error;
+	union map_info info;
+};
+
+/*
+ * For request-based dm.
+ * One of these is allocated per bio.
+ */
+struct dm_rq_clone_bio_info {
+	struct bio *orig;
+	struct request *rq;
+};
+
 union map_info *dm_get_mapinfo(struct bio *bio)
 {
 	if (bio && bio->bi_private)
@@ -141,11 +167,16 @@ struct mapped_device {
 
 	/* forced geometry settings */
 	struct hd_geometry geometry;
+
+	/* sysfs handle */
+	struct kobject kobj;
 };
 
 #define MIN_IOS 256
 static struct kmem_cache *_io_cache;
 static struct kmem_cache *_tio_cache;
+static struct kmem_cache *_rq_tio_cache;
+static struct kmem_cache *_rq_bio_info_cache;
 
 static int __init local_init(void)
 {
@@ -161,9 +192,17 @@ static int __init local_init(void)
 	if (!_tio_cache)
 		goto out_free_io_cache;
 
+	_rq_tio_cache = KMEM_CACHE(dm_rq_target_io, 0);
+	if (!_rq_tio_cache)
+		goto out_free_tio_cache;
+
+	_rq_bio_info_cache = KMEM_CACHE(dm_rq_clone_bio_info, 0);
+	if (!_rq_bio_info_cache)
+		goto out_free_rq_tio_cache;
+
 	r = dm_uevent_init();
 	if (r)
-		goto out_free_tio_cache;
+		goto out_free_rq_bio_info_cache;
 
 	_major = major;
 	r = register_blkdev(_major, _name);
@@ -177,6 +216,10 @@ static int __init local_init(void)
 
 out_uevent_exit:
 	dm_uevent_exit();
+out_free_rq_bio_info_cache:
+	kmem_cache_destroy(_rq_bio_info_cache);
+out_free_rq_tio_cache:
+	kmem_cache_destroy(_rq_tio_cache);
 out_free_tio_cache:
 	kmem_cache_destroy(_tio_cache);
 out_free_io_cache:
@@ -187,6 +230,8 @@ out_free_io_cache:
 
 static void local_exit(void)
 {
+	kmem_cache_destroy(_rq_bio_info_cache);
+	kmem_cache_destroy(_rq_tio_cache);
 	kmem_cache_destroy(_tio_cache);
 	kmem_cache_destroy(_io_cache);
 	unregister_blkdev(_major, _name);
@@ -375,7 +420,7 @@ static void start_io_acct(struct dm_io *io)
 	dm_disk(md)->part0.in_flight = atomic_inc_return(&md->pending);
 }
 
-static int end_io_acct(struct dm_io *io)
+static void end_io_acct(struct dm_io *io)
 {
 	struct mapped_device *md = io->md;
 	struct bio *bio = io->bio;
@@ -391,7 +436,9 @@ static int end_io_acct(struct dm_io *io)
 	dm_disk(md)->part0.in_flight = pending =
 		atomic_dec_return(&md->pending);
 
-	return !pending;
+	/* nudge anyone waiting on suspend queue */
+	if (!pending)
+		wake_up(&md->wait);
 }
 
 /*
@@ -499,13 +546,10 @@ static void dec_pending(struct dm_io *io, int error)
 			spin_unlock_irqrestore(&io->md->pushback_lock, flags);
 		}
 
-		if (end_io_acct(io))
-			/* nudge anyone waiting on suspend queue */
-			wake_up(&io->md->wait);
+		end_io_acct(io);
 
 		if (io->error != DM_ENDIO_REQUEUE) {
-			blk_add_trace_bio(io->md->queue, io->bio,
-					  BLK_TA_COMPLETE);
+			trace_block_bio_complete(io->md->queue, io->bio);
 
 			bio_endio(io->bio, io->error);
 		}
@@ -598,7 +642,7 @@ static void __map_bio(struct dm_target *ti, struct bio *clone,
 	if (r == DM_MAPIO_REMAPPED) {
 		/* the bio has been remapped so dispatch it */
 
-		blk_add_trace_remap(bdev_get_queue(clone->bi_bdev), clone,
+		trace_block_remap(bdev_get_queue(clone->bi_bdev), clone,
 				    tio->io->bio->bi_bdev->bd_dev,
 				    clone->bi_sector, sector);
 
@@ -794,7 +838,11 @@ static int __split_bio(struct mapped_device *md, struct bio *bio)
 	ci.map = dm_get_table(md);
 	if (unlikely(!ci.map))
 		return -EIO;
-
+	if (unlikely(bio_barrier(bio) && !dm_table_barrier_ok(ci.map))) {
+		dm_table_put(ci.map);
+		bio_endio(bio, -EOPNOTSUPP);
+		return 0;
+	}
 	ci.md = md;
 	ci.bio = bio;
 	ci.io = alloc_io(md);
@@ -878,15 +926,6 @@ static int dm_request(struct request_queue *q, struct bio *bio)
 	struct mapped_device *md = q->queuedata;
 	int cpu;
 
-	/*
-	 * There is no use in forwarding any barrier request since we can't
-	 * guarantee it is (or can be) handled by the targets correctly.
-	 */
-	if (unlikely(bio_barrier(bio))) {
-		bio_endio(bio, -EOPNOTSUPP);
-		return 0;
-	}
-
 	down_read(&md->io_lock);
 
 	cpu = part_stat_lock();
@@ -937,16 +976,18 @@ static void dm_unplug_all(struct request_queue *q)
 
 static int dm_any_congested(void *congested_data, int bdi_bits)
 {
-	int r;
-	struct mapped_device *md = (struct mapped_device *) congested_data;
-	struct dm_table *map = dm_get_table(md);
+	int r = bdi_bits;
+	struct mapped_device *md = congested_data;
+	struct dm_table *map;
 
-	if (!map || test_bit(DMF_BLOCK_IO, &md->flags))
-		r = bdi_bits;
-	else
-		r = dm_table_any_congested(map, bdi_bits);
+	if (!test_bit(DMF_BLOCK_IO, &md->flags)) {
+		map = dm_get_table(md);
+		if (map) {
+			r = dm_table_any_congested(map, bdi_bits);
+			dm_table_put(map);
+		}
+	}
 
-	dm_table_put(map);
 	return r;
 }
 
@@ -1083,7 +1124,7 @@ static struct mapped_device *alloc_dev(int minor)
 	if (!md->tio_pool)
 		goto bad_tio_pool;
 
-	md->bs = bioset_create(16, 16);
+	md->bs = bioset_create(16, 0);
 	if (!md->bs)
 		goto bad_no_bioset;
 
@@ -1206,10 +1247,12 @@ static int __bind(struct mapped_device *md, struct dm_table *t)
 
 	if (md->suspended_bdev)
 		__set_size(md, size);
-	if (size == 0)
+
+	if (!size) {
+		dm_table_destroy(t);
 		return 0;
+	}
 
-	dm_table_get(t);
 	dm_table_event_callback(t, event_callback, md);
 
 	write_lock(&md->map_lock);
@@ -1231,7 +1274,7 @@ static void __unbind(struct mapped_device *md)
 	write_lock(&md->map_lock);
 	md->map = NULL;
 	write_unlock(&md->map_lock);
-	dm_table_put(map);
+	dm_table_destroy(map);
 }
 
 /*
@@ -1245,6 +1288,8 @@ int dm_create(int minor, struct mapped_device **result)
 	if (!md)
 		return -ENXIO;
 
+	dm_sysfs_init(md);
+
 	*result = md;
 	return 0;
 }
@@ -1320,8 +1365,9 @@ void dm_put(struct mapped_device *md)
 			dm_table_presuspend_targets(map);
 			dm_table_postsuspend_targets(map);
 		}
-		__unbind(md);
+		dm_sysfs_exit(md);
 		dm_table_put(map);
+		__unbind(md);
 		free_dev(md);
 	}
 }
@@ -1659,6 +1705,27 @@ struct gendisk *dm_disk(struct mapped_device *md)
 	return md->disk;
 }
 
+struct kobject *dm_kobject(struct mapped_device *md)
+{
+	return &md->kobj;
+}
+
+/*
+ * struct mapped_device should not be exported outside of dm.c
+ * so use this check to verify that kobj is part of md structure
+ */
+struct mapped_device *dm_get_from_kobject(struct kobject *kobj)
+{
+	struct mapped_device *md;
+
+	md = container_of(kobj, struct mapped_device, kobj);
+	if (&md->kobj != kobj)
+		return NULL;
+
+	dm_get(md);
+	return md;
+}
+
 int dm_suspended(struct mapped_device *md)
 {
 	return test_bit(DMF_SUSPENDED, &md->flags);
diff --git a/drivers/md/dm.h b/drivers/md/dm.h
index 0ade60cdef4..20194e000c5 100644
--- a/drivers/md/dm.h
+++ b/drivers/md/dm.h
@@ -36,6 +36,7 @@ struct dm_table;
 /*-----------------------------------------------------------------
  * Internal table functions.
  *---------------------------------------------------------------*/
+void dm_table_destroy(struct dm_table *t);
 void dm_table_event_callback(struct dm_table *t,
 			     void (*fn)(void *), void *context);
 struct dm_target *dm_table_get_target(struct dm_table *t, unsigned int index);
@@ -51,6 +52,7 @@ int dm_table_any_congested(struct dm_table *t, int bdi_bits);
  * To check the return value from dm_table_find_target().
  */
 #define dm_target_is_valid(t) ((t)->table)
+int dm_table_barrier_ok(struct dm_table *t);
 
 /*-----------------------------------------------------------------
  * A registry of target types.
@@ -72,6 +74,14 @@ int dm_interface_init(void);
 void dm_interface_exit(void);
 
 /*
+ * sysfs interface
+ */
+int dm_sysfs_init(struct mapped_device *md);
+void dm_sysfs_exit(struct mapped_device *md);
+struct kobject *dm_kobject(struct mapped_device *md);
+struct mapped_device *dm_get_from_kobject(struct kobject *kobj);
+
+/*
  * Targets for linear and striped mappings
  */
 int dm_linear_init(void);
diff --git a/drivers/md/faulty.c b/drivers/md/faulty.c
index f26c1f9a475..86d9adf90e7 100644
--- a/drivers/md/faulty.c
+++ b/drivers/md/faulty.c
@@ -283,7 +283,6 @@ static int reconfig(mddev_t *mddev, int layout, int chunk_size)
 static int run(mddev_t *mddev)
 {
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 	int i;
 
 	conf_t *conf = kmalloc(sizeof(*conf), GFP_KERNEL);
@@ -296,7 +295,7 @@ static int run(mddev_t *mddev)
 	}
 	conf->nfaults = 0;
 
-	rdev_for_each(rdev, tmp, mddev)
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		conf->rdev = rdev;
 
 	mddev->array_sectors = mddev->size * 2;
diff --git a/drivers/md/linear.c b/drivers/md/linear.c
index 3b90c5c924e..1e3aea9eecf 100644
--- a/drivers/md/linear.c
+++ b/drivers/md/linear.c
@@ -105,7 +105,6 @@ static linear_conf_t *linear_conf(mddev_t *mddev, int raid_disks)
 	int i, nb_zone, cnt;
 	sector_t min_sectors;
 	sector_t curr_sector;
-	struct list_head *tmp;
 
 	conf = kzalloc (sizeof (*conf) + raid_disks*sizeof(dev_info_t),
 			GFP_KERNEL);
@@ -115,7 +114,7 @@ static linear_conf_t *linear_conf(mddev_t *mddev, int raid_disks)
 	cnt = 0;
 	conf->array_sectors = 0;
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		int j = rdev->raid_disk;
 		dev_info_t *disk = conf->disks + j;
 
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 1b1d32694f6..41e2509bf89 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -214,20 +214,33 @@ static inline mddev_t *mddev_get(mddev_t *mddev)
 	return mddev;
 }
 
+static void mddev_delayed_delete(struct work_struct *ws)
+{
+	mddev_t *mddev = container_of(ws, mddev_t, del_work);
+	kobject_del(&mddev->kobj);
+	kobject_put(&mddev->kobj);
+}
+
 static void mddev_put(mddev_t *mddev)
 {
 	if (!atomic_dec_and_lock(&mddev->active, &all_mddevs_lock))
 		return;
-	if (!mddev->raid_disks && list_empty(&mddev->disks)) {
+	if (!mddev->raid_disks && list_empty(&mddev->disks) &&
+	    !mddev->hold_active) {
 		list_del(&mddev->all_mddevs);
-		spin_unlock(&all_mddevs_lock);
-		blk_cleanup_queue(mddev->queue);
-		if (mddev->sysfs_state)
-			sysfs_put(mddev->sysfs_state);
-		mddev->sysfs_state = NULL;
-		kobject_put(&mddev->kobj);
-	} else
-		spin_unlock(&all_mddevs_lock);
+		if (mddev->gendisk) {
+			/* we did a probe so need to clean up.
+			 * Call schedule_work inside the spinlock
+			 * so that flush_scheduled_work() after
+			 * mddev_find will succeed in waiting for the
+			 * work to be done.
+			 */
+			INIT_WORK(&mddev->del_work, mddev_delayed_delete);
+			schedule_work(&mddev->del_work);
+		} else
+			kfree(mddev);
+	}
+	spin_unlock(&all_mddevs_lock);
 }
 
 static mddev_t * mddev_find(dev_t unit)
@@ -236,15 +249,50 @@ static mddev_t * mddev_find(dev_t unit)
 
  retry:
 	spin_lock(&all_mddevs_lock);
-	list_for_each_entry(mddev, &all_mddevs, all_mddevs)
-		if (mddev->unit == unit) {
-			mddev_get(mddev);
+
+	if (unit) {
+		list_for_each_entry(mddev, &all_mddevs, all_mddevs)
+			if (mddev->unit == unit) {
+				mddev_get(mddev);
+				spin_unlock(&all_mddevs_lock);
+				kfree(new);
+				return mddev;
+			}
+
+		if (new) {
+			list_add(&new->all_mddevs, &all_mddevs);
 			spin_unlock(&all_mddevs_lock);
-			kfree(new);
-			return mddev;
+			new->hold_active = UNTIL_IOCTL;
+			return new;
 		}
-
-	if (new) {
+	} else if (new) {
+		/* find an unused unit number */
+		static int next_minor = 512;
+		int start = next_minor;
+		int is_free = 0;
+		int dev = 0;
+		while (!is_free) {
+			dev = MKDEV(MD_MAJOR, next_minor);
+			next_minor++;
+			if (next_minor > MINORMASK)
+				next_minor = 0;
+			if (next_minor == start) {
+				/* Oh dear, all in use. */
+				spin_unlock(&all_mddevs_lock);
+				kfree(new);
+				return NULL;
+			}
+				
+			is_free = 1;
+			list_for_each_entry(mddev, &all_mddevs, all_mddevs)
+				if (mddev->unit == dev) {
+					is_free = 0;
+					break;
+				}
+		}
+		new->unit = dev;
+		new->md_minor = MINOR(dev);
+		new->hold_active = UNTIL_STOP;
 		list_add(&new->all_mddevs, &all_mddevs);
 		spin_unlock(&all_mddevs_lock);
 		return new;
@@ -275,16 +323,6 @@ static mddev_t * mddev_find(dev_t unit)
 	new->resync_max = MaxSector;
 	new->level = LEVEL_NONE;
 
-	new->queue = blk_alloc_queue(GFP_KERNEL);
-	if (!new->queue) {
-		kfree(new);
-		return NULL;
-	}
-	/* Can be unlocked because the queue is new: no concurrency */
-	queue_flag_set_unlocked(QUEUE_FLAG_CLUSTER, new->queue);
-
-	blk_queue_make_request(new->queue, md_fail_request);
-
 	goto retry;
 }
 
@@ -307,25 +345,23 @@ static inline void mddev_unlock(mddev_t * mddev)
 
 static mdk_rdev_t * find_rdev_nr(mddev_t *mddev, int nr)
 {
-	mdk_rdev_t * rdev;
-	struct list_head *tmp;
+	mdk_rdev_t *rdev;
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		if (rdev->desc_nr == nr)
 			return rdev;
-	}
+
 	return NULL;
 }
 
 static mdk_rdev_t * find_rdev(mddev_t * mddev, dev_t dev)
 {
-	struct list_head *tmp;
 	mdk_rdev_t *rdev;
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		if (rdev->bdev->bd_dev == dev)
 			return rdev;
-	}
+
 	return NULL;
 }
 
@@ -861,7 +897,6 @@ static int super_90_validate(mddev_t *mddev, mdk_rdev_t *rdev)
 static void super_90_sync(mddev_t *mddev, mdk_rdev_t *rdev)
 {
 	mdp_super_t *sb;
-	struct list_head *tmp;
 	mdk_rdev_t *rdev2;
 	int next_spare = mddev->raid_disks;
 
@@ -933,7 +968,7 @@ static void super_90_sync(mddev_t *mddev, mdk_rdev_t *rdev)
 		sb->state |= (1<<MD_SB_BITMAP_PRESENT);
 
 	sb->disks[0].state = (1<<MD_DISK_REMOVED);
-	rdev_for_each(rdev2, tmp, mddev) {
+	list_for_each_entry(rdev2, &mddev->disks, same_set) {
 		mdp_disk_t *d;
 		int desc_nr;
 		if (rdev2->raid_disk >= 0 && test_bit(In_sync, &rdev2->flags)
@@ -1259,7 +1294,6 @@ static int super_1_validate(mddev_t *mddev, mdk_rdev_t *rdev)
 static void super_1_sync(mddev_t *mddev, mdk_rdev_t *rdev)
 {
 	struct mdp_superblock_1 *sb;
-	struct list_head *tmp;
 	mdk_rdev_t *rdev2;
 	int max_dev, i;
 	/* make rdev->sb match mddev and rdev data. */
@@ -1307,7 +1341,7 @@ static void super_1_sync(mddev_t *mddev, mdk_rdev_t *rdev)
 	}
 
 	max_dev = 0;
-	rdev_for_each(rdev2, tmp, mddev)
+	list_for_each_entry(rdev2, &mddev->disks, same_set)
 		if (rdev2->desc_nr+1 > max_dev)
 			max_dev = rdev2->desc_nr+1;
 
@@ -1316,7 +1350,7 @@ static void super_1_sync(mddev_t *mddev, mdk_rdev_t *rdev)
 	for (i=0; i<max_dev;i++)
 		sb->dev_roles[i] = cpu_to_le16(0xfffe);
 	
-	rdev_for_each(rdev2, tmp, mddev) {
+	list_for_each_entry(rdev2, &mddev->disks, same_set) {
 		i = rdev2->desc_nr;
 		if (test_bit(Faulty, &rdev2->flags))
 			sb->dev_roles[i] = cpu_to_le16(0xfffe);
@@ -1466,6 +1500,9 @@ static int bind_rdev_to_array(mdk_rdev_t * rdev, mddev_t * mddev)
 
 	list_add_rcu(&rdev->same_set, &mddev->disks);
 	bd_claim_by_disk(rdev->bdev, rdev->bdev->bd_holder, mddev->gendisk);
+
+	/* May as well allow recovery to be retried once */
+	mddev->recovery_disabled = 0;
 	return 0;
 
  fail:
@@ -1571,8 +1608,7 @@ static void kick_rdev_from_array(mdk_rdev_t * rdev)
 
 static void export_array(mddev_t *mddev)
 {
-	struct list_head *tmp;
-	mdk_rdev_t *rdev;
+	mdk_rdev_t *rdev, *tmp;
 
 	rdev_for_each(rdev, tmp, mddev) {
 		if (!rdev->mddev) {
@@ -1593,7 +1629,7 @@ static void print_desc(mdp_disk_t *desc)
 		desc->major,desc->minor,desc->raid_disk,desc->state);
 }
 
-static void print_sb(mdp_super_t *sb)
+static void print_sb_90(mdp_super_t *sb)
 {
 	int i;
 
@@ -1624,10 +1660,57 @@ static void print_sb(mdp_super_t *sb)
 	}
 	printk(KERN_INFO "md:     THIS: ");
 	print_desc(&sb->this_disk);
-
 }
 
-static void print_rdev(mdk_rdev_t *rdev)
+static void print_sb_1(struct mdp_superblock_1 *sb)
+{
+	__u8 *uuid;
+
+	uuid = sb->set_uuid;
+	printk(KERN_INFO "md:  SB: (V:%u) (F:0x%08x) Array-ID:<%02x%02x%02x%02x"
+			":%02x%02x:%02x%02x:%02x%02x:%02x%02x%02x%02x%02x%02x>\n"
+	       KERN_INFO "md:    Name: \"%s\" CT:%llu\n",
+		le32_to_cpu(sb->major_version),
+		le32_to_cpu(sb->feature_map),
+		uuid[0], uuid[1], uuid[2], uuid[3],
+		uuid[4], uuid[5], uuid[6], uuid[7],
+		uuid[8], uuid[9], uuid[10], uuid[11],
+		uuid[12], uuid[13], uuid[14], uuid[15],
+		sb->set_name,
+		(unsigned long long)le64_to_cpu(sb->ctime)
+		       & MD_SUPERBLOCK_1_TIME_SEC_MASK);
+
+	uuid = sb->device_uuid;
+	printk(KERN_INFO "md:       L%u SZ%llu RD:%u LO:%u CS:%u DO:%llu DS:%llu SO:%llu"
+			" RO:%llu\n"
+	       KERN_INFO "md:     Dev:%08x UUID: %02x%02x%02x%02x:%02x%02x:%02x%02x:%02x%02x"
+			":%02x%02x%02x%02x%02x%02x\n"
+	       KERN_INFO "md:       (F:0x%08x) UT:%llu Events:%llu ResyncOffset:%llu CSUM:0x%08x\n"
+	       KERN_INFO "md:         (MaxDev:%u) \n",
+		le32_to_cpu(sb->level),
+		(unsigned long long)le64_to_cpu(sb->size),
+		le32_to_cpu(sb->raid_disks),
+		le32_to_cpu(sb->layout),
+		le32_to_cpu(sb->chunksize),
+		(unsigned long long)le64_to_cpu(sb->data_offset),
+		(unsigned long long)le64_to_cpu(sb->data_size),
+		(unsigned long long)le64_to_cpu(sb->super_offset),
+		(unsigned long long)le64_to_cpu(sb->recovery_offset),
+		le32_to_cpu(sb->dev_number),
+		uuid[0], uuid[1], uuid[2], uuid[3],
+		uuid[4], uuid[5], uuid[6], uuid[7],
+		uuid[8], uuid[9], uuid[10], uuid[11],
+		uuid[12], uuid[13], uuid[14], uuid[15],
+		sb->devflags,
+		(unsigned long long)le64_to_cpu(sb->utime) & MD_SUPERBLOCK_1_TIME_SEC_MASK,
+		(unsigned long long)le64_to_cpu(sb->events),
+		(unsigned long long)le64_to_cpu(sb->resync_offset),
+		le32_to_cpu(sb->sb_csum),
+		le32_to_cpu(sb->max_dev)
+		);
+}
+
+static void print_rdev(mdk_rdev_t *rdev, int major_version)
 {
 	char b[BDEVNAME_SIZE];
 	printk(KERN_INFO "md: rdev %s, SZ:%08llu F:%d S:%d DN:%u\n",
@@ -1635,15 +1718,22 @@ static void print_rdev(mdk_rdev_t *rdev)
 	        test_bit(Faulty, &rdev->flags), test_bit(In_sync, &rdev->flags),
 	        rdev->desc_nr);
 	if (rdev->sb_loaded) {
-		printk(KERN_INFO "md: rdev superblock:\n");
-		print_sb((mdp_super_t*)page_address(rdev->sb_page));
+		printk(KERN_INFO "md: rdev superblock (MJ:%d):\n", major_version);
+		switch (major_version) {
+		case 0:
+			print_sb_90((mdp_super_t*)page_address(rdev->sb_page));
+			break;
+		case 1:
+			print_sb_1((struct mdp_superblock_1 *)page_address(rdev->sb_page));
+			break;
+		}
 	} else
 		printk(KERN_INFO "md: no rdev superblock!\n");
 }
 
 static void md_print_devices(void)
 {
-	struct list_head *tmp, *tmp2;
+	struct list_head *tmp;
 	mdk_rdev_t *rdev;
 	mddev_t *mddev;
 	char b[BDEVNAME_SIZE];
@@ -1658,12 +1748,12 @@ static void md_print_devices(void)
 			bitmap_print_sb(mddev->bitmap);
 		else
 			printk("%s: ", mdname(mddev));
-		rdev_for_each(rdev, tmp2, mddev)
+		list_for_each_entry(rdev, &mddev->disks, same_set)
 			printk("<%s>", bdevname(rdev->bdev,b));
 		printk("\n");
 
-		rdev_for_each(rdev, tmp2, mddev)
-			print_rdev(rdev);
+		list_for_each_entry(rdev, &mddev->disks, same_set)
+			print_rdev(rdev, mddev->major_version);
 	}
 	printk("md:	**********************************\n");
 	printk("\n");
@@ -1679,9 +1769,8 @@ static void sync_sbs(mddev_t * mddev, int nospares)
 	 * with the rest of the array)
 	 */
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		if (rdev->sb_events == mddev->events ||
 		    (nospares &&
 		     rdev->raid_disk < 0 &&
@@ -1699,7 +1788,6 @@ static void sync_sbs(mddev_t * mddev, int nospares)
 
 static void md_update_sb(mddev_t * mddev, int force_change)
 {
-	struct list_head *tmp;
 	mdk_rdev_t *rdev;
 	int sync_req;
 	int nospares = 0;
@@ -1790,7 +1878,7 @@ repeat:
 		mdname(mddev),mddev->in_sync);
 
 	bitmap_update_sb(mddev->bitmap);
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		char b[BDEVNAME_SIZE];
 		dprintk(KERN_INFO "md: ");
 		if (rdev->sb_loaded != 1)
@@ -1999,7 +2087,6 @@ slot_store(mdk_rdev_t *rdev, const char *buf, size_t len)
 		md_wakeup_thread(rdev->mddev->thread);
 	} else if (rdev->mddev->pers) {
 		mdk_rdev_t *rdev2;
-		struct list_head *tmp;
 		/* Activating a spare .. or possibly reactivating
 		 * if we every get bitmaps working here.
 		 */
@@ -2010,7 +2097,7 @@ slot_store(mdk_rdev_t *rdev, const char *buf, size_t len)
 		if (rdev->mddev->pers->hot_add_disk == NULL)
 			return -EINVAL;
 
-		rdev_for_each(rdev2, tmp, rdev->mddev)
+		list_for_each_entry(rdev2, &rdev->mddev->disks, same_set)
 			if (rdev2->raid_disk == slot)
 				return -EEXIST;
 
@@ -2125,14 +2212,14 @@ rdev_size_store(mdk_rdev_t *rdev, const char *buf, size_t len)
 		 */
 		mddev_t *mddev;
 		int overlap = 0;
-		struct list_head *tmp, *tmp2;
+		struct list_head *tmp;
 
 		mddev_unlock(my_mddev);
 		for_each_mddev(mddev, tmp) {
 			mdk_rdev_t *rdev2;
 
 			mddev_lock(mddev);
-			rdev_for_each(rdev2, tmp2, mddev)
+			list_for_each_entry(rdev2, &mddev->disks, same_set)
 				if (test_bit(AllReserved, &rdev2->flags) ||
 				    (rdev->bdev == rdev2->bdev &&
 				     rdev != rdev2 &&
@@ -2328,8 +2415,7 @@ abort_free:
 static void analyze_sbs(mddev_t * mddev)
 {
 	int i;
-	struct list_head *tmp;
-	mdk_rdev_t *rdev, *freshest;
+	mdk_rdev_t *rdev, *freshest, *tmp;
 	char b[BDEVNAME_SIZE];
 
 	freshest = NULL;
@@ -3046,7 +3132,7 @@ action_store(mddev_t *mddev, const char *page, size_t len)
 	}
 	set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 	md_wakeup_thread(mddev->thread);
-	sysfs_notify(&mddev->kobj, NULL, "sync_action");
+	sysfs_notify_dirent(mddev->sysfs_action);
 	return len;
 }
 
@@ -3404,6 +3490,8 @@ md_attr_store(struct kobject *kobj, struct attribute *attr,
 	if (!capable(CAP_SYS_ADMIN))
 		return -EACCES;
 	rv = mddev_lock(mddev);
+	if (mddev->hold_active == UNTIL_IOCTL)
+		mddev->hold_active = 0;
 	if (!rv) {
 		rv = entry->store(mddev, page, length);
 		mddev_unlock(mddev);
@@ -3414,6 +3502,17 @@ md_attr_store(struct kobject *kobj, struct attribute *attr,
 static void md_free(struct kobject *ko)
 {
 	mddev_t *mddev = container_of(ko, mddev_t, kobj);
+
+	if (mddev->sysfs_state)
+		sysfs_put(mddev->sysfs_state);
+
+	if (mddev->gendisk) {
+		del_gendisk(mddev->gendisk);
+		put_disk(mddev->gendisk);
+	}
+	if (mddev->queue)
+		blk_cleanup_queue(mddev->queue);
+
 	kfree(mddev);
 }
 
@@ -3429,34 +3528,74 @@ static struct kobj_type md_ktype = {
 
 int mdp_major = 0;
 
-static struct kobject *md_probe(dev_t dev, int *part, void *data)
+static int md_alloc(dev_t dev, char *name)
 {
 	static DEFINE_MUTEX(disks_mutex);
 	mddev_t *mddev = mddev_find(dev);
 	struct gendisk *disk;
-	int partitioned = (MAJOR(dev) != MD_MAJOR);
-	int shift = partitioned ? MdpMinorShift : 0;
-	int unit = MINOR(dev) >> shift;
+	int partitioned;
+	int shift;
+	int unit;
 	int error;
 
 	if (!mddev)
-		return NULL;
+		return -ENODEV;
+
+	partitioned = (MAJOR(mddev->unit) != MD_MAJOR);
+	shift = partitioned ? MdpMinorShift : 0;
+	unit = MINOR(mddev->unit) >> shift;
+
+	/* wait for any previous instance if this device
+	 * to be completed removed (mddev_delayed_delete).
+	 */
+	flush_scheduled_work();
 
 	mutex_lock(&disks_mutex);
 	if (mddev->gendisk) {
 		mutex_unlock(&disks_mutex);
 		mddev_put(mddev);
-		return NULL;
+		return -EEXIST;
+	}
+
+	if (name) {
+		/* Need to ensure that 'name' is not a duplicate.
+		 */
+		mddev_t *mddev2;
+		spin_lock(&all_mddevs_lock);
+
+		list_for_each_entry(mddev2, &all_mddevs, all_mddevs)
+			if (mddev2->gendisk &&
+			    strcmp(mddev2->gendisk->disk_name, name) == 0) {
+				spin_unlock(&all_mddevs_lock);
+				return -EEXIST;
+			}
+		spin_unlock(&all_mddevs_lock);
+	}
+
+	mddev->queue = blk_alloc_queue(GFP_KERNEL);
+	if (!mddev->queue) {
+		mutex_unlock(&disks_mutex);
+		mddev_put(mddev);
+		return -ENOMEM;
 	}
+	/* Can be unlocked because the queue is new: no concurrency */
+	queue_flag_set_unlocked(QUEUE_FLAG_CLUSTER, mddev->queue);
+
+	blk_queue_make_request(mddev->queue, md_fail_request);
+
 	disk = alloc_disk(1 << shift);
 	if (!disk) {
 		mutex_unlock(&disks_mutex);
+		blk_cleanup_queue(mddev->queue);
+		mddev->queue = NULL;
 		mddev_put(mddev);
-		return NULL;
+		return -ENOMEM;
 	}
-	disk->major = MAJOR(dev);
+	disk->major = MAJOR(mddev->unit);
 	disk->first_minor = unit << shift;
-	if (partitioned)
+	if (name)
+		strcpy(disk->disk_name, name);
+	else if (partitioned)
 		sprintf(disk->disk_name, "md_d%d", unit);
 	else
 		sprintf(disk->disk_name, "md%d", unit);
@@ -3464,7 +3603,7 @@ static struct kobject *md_probe(dev_t dev, int *part, void *data)
 	disk->private_data = mddev;
 	disk->queue = mddev->queue;
 	/* Allow extended partitions.  This makes the
-	 * 'mdp' device redundant, but we can really
+	 * 'mdp' device redundant, but we can't really
 	 * remove it now.
 	 */
 	disk->flags |= GENHD_FL_EXT_DEVT;
@@ -3480,9 +3619,35 @@ static struct kobject *md_probe(dev_t dev, int *part, void *data)
 		kobject_uevent(&mddev->kobj, KOBJ_ADD);
 		mddev->sysfs_state = sysfs_get_dirent(mddev->kobj.sd, "array_state");
 	}
+	mddev_put(mddev);
+	return 0;
+}
+
+static struct kobject *md_probe(dev_t dev, int *part, void *data)
+{
+	md_alloc(dev, NULL);
 	return NULL;
 }
 
+static int add_named_array(const char *val, struct kernel_param *kp)
+{
+	/* val must be "md_*" where * is not all digits.
+	 * We allocate an array with a large free minor number, and
+	 * set the name to val.  val must not already be an active name.
+	 */
+	int len = strlen(val);
+	char buf[DISK_NAME_LEN];
+
+	while (len && val[len-1] == '\n')
+		len--;
+	if (len >= DISK_NAME_LEN)
+		return -E2BIG;
+	strlcpy(buf, val, len+1);
+	if (strncmp(buf, "md_", 3) != 0)
+		return -EINVAL;
+	return md_alloc(0, buf);
+}
+
 static void md_safemode_timeout(unsigned long data)
 {
 	mddev_t *mddev = (mddev_t *) data;
@@ -3501,7 +3666,6 @@ static int do_md_run(mddev_t * mddev)
 {
 	int err;
 	int chunk_size;
-	struct list_head *tmp;
 	mdk_rdev_t *rdev;
 	struct gendisk *disk;
 	struct mdk_personality *pers;
@@ -3540,7 +3704,7 @@ static int do_md_run(mddev_t * mddev)
 		}
 
 		/* devices must have minimum size of one chunk */
-		rdev_for_each(rdev, tmp, mddev) {
+		list_for_each_entry(rdev, &mddev->disks, same_set) {
 			if (test_bit(Faulty, &rdev->flags))
 				continue;
 			if (rdev->size < chunk_size / 1024) {
@@ -3565,7 +3729,7 @@ static int do_md_run(mddev_t * mddev)
 	 * the only valid external interface is through the md
 	 * device.
 	 */
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		if (test_bit(Faulty, &rdev->flags))
 			continue;
 		sync_blockdev(rdev->bdev);
@@ -3630,10 +3794,10 @@ static int do_md_run(mddev_t * mddev)
 		 */
 		char b[BDEVNAME_SIZE], b2[BDEVNAME_SIZE];
 		mdk_rdev_t *rdev2;
-		struct list_head *tmp2;
 		int warned = 0;
-		rdev_for_each(rdev, tmp, mddev) {
-			rdev_for_each(rdev2, tmp2, mddev) {
+
+		list_for_each_entry(rdev, &mddev->disks, same_set)
+			list_for_each_entry(rdev2, &mddev->disks, same_set) {
 				if (rdev < rdev2 &&
 				    rdev->bdev->bd_contains ==
 				    rdev2->bdev->bd_contains) {
@@ -3647,7 +3811,7 @@ static int do_md_run(mddev_t * mddev)
 					warned = 1;
 				}
 			}
-		}
+
 		if (warned)
 			printk(KERN_WARNING
 			       "True protection against single-disk"
@@ -3684,6 +3848,7 @@ static int do_md_run(mddev_t * mddev)
 			printk(KERN_WARNING
 			       "md: cannot register extra attributes for %s\n",
 			       mdname(mddev));
+		mddev->sysfs_action = sysfs_get_dirent(mddev->kobj.sd, "sync_action");
 	} else if (mddev->ro == 2) /* auto-readonly not meaningful */
 		mddev->ro = 0;
 
@@ -3694,7 +3859,7 @@ static int do_md_run(mddev_t * mddev)
 	mddev->safemode_delay = (200 * HZ)/1000 +1; /* 200 msec delay */
 	mddev->in_sync = 1;
 
-	rdev_for_each(rdev, tmp, mddev)
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		if (rdev->raid_disk >= 0) {
 			char nm[20];
 			sprintf(nm, "rd%d", rdev->raid_disk);
@@ -3725,9 +3890,8 @@ static int do_md_run(mddev_t * mddev)
 	 * it will remove the drives and not do the right thing
 	 */
 	if (mddev->degraded && !mddev->sync_thread) {
-		struct list_head *rtmp;
 		int spares = 0;
-		rdev_for_each(rdev, rtmp, mddev)
+		list_for_each_entry(rdev, &mddev->disks, same_set)
 			if (rdev->raid_disk >= 0 &&
 			    !test_bit(In_sync, &rdev->flags) &&
 			    !test_bit(Faulty, &rdev->flags))
@@ -3754,7 +3918,8 @@ static int do_md_run(mddev_t * mddev)
 	mddev->changed = 1;
 	md_new_event(mddev);
 	sysfs_notify_dirent(mddev->sysfs_state);
-	sysfs_notify(&mddev->kobj, NULL, "sync_action");
+	if (mddev->sysfs_action)
+		sysfs_notify_dirent(mddev->sysfs_action);
 	sysfs_notify(&mddev->kobj, NULL, "degraded");
 	kobject_uevent(&disk_to_dev(mddev->gendisk)->kobj, KOBJ_CHANGE);
 	return 0;
@@ -3854,9 +4019,12 @@ static int do_md_stop(mddev_t * mddev, int mode, int is_open)
 			mddev->queue->merge_bvec_fn = NULL;
 			mddev->queue->unplug_fn = NULL;
 			mddev->queue->backing_dev_info.congested_fn = NULL;
-			if (mddev->pers->sync_request)
+			if (mddev->pers->sync_request) {
 				sysfs_remove_group(&mddev->kobj, &md_redundancy_group);
-
+				if (mddev->sysfs_action)
+					sysfs_put(mddev->sysfs_action);
+				mddev->sysfs_action = NULL;
+			}
 			module_put(mddev->pers->owner);
 			mddev->pers = NULL;
 			/* tell userspace to handle 'inactive' */
@@ -3883,7 +4051,6 @@ static int do_md_stop(mddev_t * mddev, int mode, int is_open)
 	 */
 	if (mode == 0) {
 		mdk_rdev_t *rdev;
-		struct list_head *tmp;
 
 		printk(KERN_INFO "md: %s stopped.\n", mdname(mddev));
 
@@ -3895,7 +4062,7 @@ static int do_md_stop(mddev_t * mddev, int mode, int is_open)
 		}
 		mddev->bitmap_offset = 0;
 
-		rdev_for_each(rdev, tmp, mddev)
+		list_for_each_entry(rdev, &mddev->disks, same_set)
 			if (rdev->raid_disk >= 0) {
 				char nm[20];
 				sprintf(nm, "rd%d", rdev->raid_disk);
@@ -3941,6 +4108,8 @@ static int do_md_stop(mddev_t * mddev, int mode, int is_open)
 		mddev->barriers_work = 0;
 		mddev->safemode = 0;
 		kobject_uevent(&disk_to_dev(mddev->gendisk)->kobj, KOBJ_CHANGE);
+		if (mddev->hold_active == UNTIL_STOP)
+			mddev->hold_active = 0;
 
 	} else if (mddev->pers)
 		printk(KERN_INFO "md: %s switched to read-only mode.\n",
@@ -3956,7 +4125,6 @@ out:
 static void autorun_array(mddev_t *mddev)
 {
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 	int err;
 
 	if (list_empty(&mddev->disks))
@@ -3964,7 +4132,7 @@ static void autorun_array(mddev_t *mddev)
 
 	printk(KERN_INFO "md: running: ");
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		char b[BDEVNAME_SIZE];
 		printk("<%s>", bdevname(rdev->bdev,b));
 	}
@@ -3991,8 +4159,7 @@ static void autorun_array(mddev_t *mddev)
  */
 static void autorun_devices(int part)
 {
-	struct list_head *tmp;
-	mdk_rdev_t *rdev0, *rdev;
+	mdk_rdev_t *rdev0, *rdev, *tmp;
 	mddev_t *mddev;
 	char b[BDEVNAME_SIZE];
 
@@ -4007,7 +4174,7 @@ static void autorun_devices(int part)
 		printk(KERN_INFO "md: considering %s ...\n",
 			bdevname(rdev0->bdev,b));
 		INIT_LIST_HEAD(&candidates);
-		rdev_for_each_list(rdev, tmp, pending_raid_disks)
+		rdev_for_each_list(rdev, tmp, &pending_raid_disks)
 			if (super_90_load(rdev, rdev0, 0) >= 0) {
 				printk(KERN_INFO "md:  adding %s ...\n",
 					bdevname(rdev->bdev,b));
@@ -4053,7 +4220,7 @@ static void autorun_devices(int part)
 		} else {
 			printk(KERN_INFO "md: created %s\n", mdname(mddev));
 			mddev->persistent = 1;
-			rdev_for_each_list(rdev, tmp, candidates) {
+			rdev_for_each_list(rdev, tmp, &candidates) {
 				list_del_init(&rdev->same_set);
 				if (bind_rdev_to_array(rdev, mddev))
 					export_rdev(rdev);
@@ -4064,7 +4231,7 @@ static void autorun_devices(int part)
 		/* on success, candidates will be empty, on error
 		 * it won't...
 		 */
-		rdev_for_each_list(rdev, tmp, candidates) {
+		rdev_for_each_list(rdev, tmp, &candidates) {
 			list_del_init(&rdev->same_set);
 			export_rdev(rdev);
 		}
@@ -4093,10 +4260,9 @@ static int get_array_info(mddev_t * mddev, void __user * arg)
 	mdu_array_info_t info;
 	int nr,working,active,failed,spare;
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 
 	nr=working=active=failed=spare=0;
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		nr++;
 		if (test_bit(Faulty, &rdev->flags))
 			failed++;
@@ -4614,9 +4780,8 @@ static int set_array_info(mddev_t * mddev, mdu_array_info_t *info)
 
 static int update_size(mddev_t *mddev, sector_t num_sectors)
 {
-	mdk_rdev_t * rdev;
+	mdk_rdev_t *rdev;
 	int rv;
-	struct list_head *tmp;
 	int fit = (num_sectors == 0);
 
 	if (mddev->pers->resize == NULL)
@@ -4638,7 +4803,7 @@ static int update_size(mddev_t *mddev, sector_t num_sectors)
 		 * grow, and re-add.
 		 */
 		return -EBUSY;
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		sector_t avail;
 		avail = rdev->size * 2;
 
@@ -5000,6 +5165,9 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 
 done_unlock:
 abort_unlock:
+	if (mddev->hold_active == UNTIL_IOCTL &&
+	    err != -EINVAL)
+		mddev->hold_active = 0;
 	mddev_unlock(mddev);
 
 	return err;
@@ -5016,14 +5184,25 @@ static int md_open(struct block_device *bdev, fmode_t mode)
 	 * Succeed if we can lock the mddev, which confirms that
 	 * it isn't being stopped right now.
 	 */
-	mddev_t *mddev = bdev->bd_disk->private_data;
+	mddev_t *mddev = mddev_find(bdev->bd_dev);
 	int err;
 
+	if (mddev->gendisk != bdev->bd_disk) {
+		/* we are racing with mddev_put which is discarding this
+		 * bd_disk.
+		 */
+		mddev_put(mddev);
+		/* Wait until bdev->bd_disk is definitely gone */
+		flush_scheduled_work();
+		/* Then retry the open from the top */
+		return -ERESTARTSYS;
+	}
+	BUG_ON(mddev != bdev->bd_disk->private_data);
+
 	if ((err = mutex_lock_interruptible_nested(&mddev->reconfig_mutex, 1)))
 		goto out;
 
 	err = 0;
-	mddev_get(mddev);
 	atomic_inc(&mddev->openers);
 	mddev_unlock(mddev);
 
@@ -5187,11 +5366,10 @@ static void status_unused(struct seq_file *seq)
 {
 	int i = 0;
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 
 	seq_printf(seq, "unused devices: ");
 
-	rdev_for_each_list(rdev, tmp, pending_raid_disks) {
+	list_for_each_entry(rdev, &pending_raid_disks, same_set) {
 		char b[BDEVNAME_SIZE];
 		i++;
 		seq_printf(seq, "%s ",
@@ -5350,7 +5528,6 @@ static int md_seq_show(struct seq_file *seq, void *v)
 {
 	mddev_t *mddev = v;
 	sector_t size;
-	struct list_head *tmp2;
 	mdk_rdev_t *rdev;
 	struct mdstat_info *mi = seq->private;
 	struct bitmap *bitmap;
@@ -5387,7 +5564,7 @@ static int md_seq_show(struct seq_file *seq, void *v)
 		}
 
 		size = 0;
-		rdev_for_each(rdev, tmp2, mddev) {
+		list_for_each_entry(rdev, &mddev->disks, same_set) {
 			char b[BDEVNAME_SIZE];
 			seq_printf(seq, " %s[%d]",
 				bdevname(rdev->bdev,b), rdev->desc_nr);
@@ -5694,7 +5871,6 @@ void md_do_sync(mddev_t *mddev)
 	struct list_head *tmp;
 	sector_t last_check;
 	int skipped = 0;
-	struct list_head *rtmp;
 	mdk_rdev_t *rdev;
 	char *desc;
 
@@ -5799,7 +5975,7 @@ void md_do_sync(mddev_t *mddev)
 		/* recovery follows the physical size of devices */
 		max_sectors = mddev->size << 1;
 		j = MaxSector;
-		rdev_for_each(rdev, rtmp, mddev)
+		list_for_each_entry(rdev, &mddev->disks, same_set)
 			if (rdev->raid_disk >= 0 &&
 			    !test_bit(Faulty, &rdev->flags) &&
 			    !test_bit(In_sync, &rdev->flags) &&
@@ -5949,7 +6125,7 @@ void md_do_sync(mddev_t *mddev)
 		} else {
 			if (!test_bit(MD_RECOVERY_INTR, &mddev->recovery))
 				mddev->curr_resync = MaxSector;
-			rdev_for_each(rdev, rtmp, mddev)
+			list_for_each_entry(rdev, &mddev->disks, same_set)
 				if (rdev->raid_disk >= 0 &&
 				    !test_bit(Faulty, &rdev->flags) &&
 				    !test_bit(In_sync, &rdev->flags) &&
@@ -5985,10 +6161,9 @@ EXPORT_SYMBOL_GPL(md_do_sync);
 static int remove_and_add_spares(mddev_t *mddev)
 {
 	mdk_rdev_t *rdev;
-	struct list_head *rtmp;
 	int spares = 0;
 
-	rdev_for_each(rdev, rtmp, mddev)
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		if (rdev->raid_disk >= 0 &&
 		    !test_bit(Blocked, &rdev->flags) &&
 		    (test_bit(Faulty, &rdev->flags) ||
@@ -6003,8 +6178,8 @@ static int remove_and_add_spares(mddev_t *mddev)
 			}
 		}
 
-	if (mddev->degraded && ! mddev->ro) {
-		rdev_for_each(rdev, rtmp, mddev) {
+	if (mddev->degraded && ! mddev->ro && !mddev->recovery_disabled) {
+		list_for_each_entry(rdev, &mddev->disks, same_set) {
 			if (rdev->raid_disk >= 0 &&
 			    !test_bit(In_sync, &rdev->flags) &&
 			    !test_bit(Blocked, &rdev->flags))
@@ -6056,7 +6231,6 @@ static int remove_and_add_spares(mddev_t *mddev)
 void md_check_recovery(mddev_t *mddev)
 {
 	mdk_rdev_t *rdev;
-	struct list_head *rtmp;
 
 
 	if (mddev->bitmap)
@@ -6120,7 +6294,7 @@ void md_check_recovery(mddev_t *mddev)
 		if (mddev->flags)
 			md_update_sb(mddev, 0);
 
-		rdev_for_each(rdev, rtmp, mddev)
+		list_for_each_entry(rdev, &mddev->disks, same_set)
 			if (test_and_clear_bit(StateChanged, &rdev->flags))
 				sysfs_notify_dirent(rdev->sysfs_state);
 
@@ -6149,13 +6323,13 @@ void md_check_recovery(mddev_t *mddev)
 			 * information must be scrapped
 			 */
 			if (!mddev->degraded)
-				rdev_for_each(rdev, rtmp, mddev)
+				list_for_each_entry(rdev, &mddev->disks, same_set)
 					rdev->saved_raid_disk = -1;
 
 			mddev->recovery = 0;
 			/* flag recovery needed just to double check */
 			set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
-			sysfs_notify(&mddev->kobj, NULL, "sync_action");
+			sysfs_notify_dirent(mddev->sysfs_action);
 			md_new_event(mddev);
 			goto unlock;
 		}
@@ -6216,7 +6390,7 @@ void md_check_recovery(mddev_t *mddev)
 				mddev->recovery = 0;
 			} else
 				md_wakeup_thread(mddev->sync_thread);
-			sysfs_notify(&mddev->kobj, NULL, "sync_action");
+			sysfs_notify_dirent(mddev->sysfs_action);
 			md_new_event(mddev);
 		}
 	unlock:
@@ -6224,7 +6398,8 @@ void md_check_recovery(mddev_t *mddev)
 			clear_bit(MD_RECOVERY_RUNNING, &mddev->recovery);
 			if (test_and_clear_bit(MD_RECOVERY_RECOVER,
 					       &mddev->recovery))
-				sysfs_notify(&mddev->kobj, NULL, "sync_action");
+				if (mddev->sysfs_action)
+					sysfs_notify_dirent(mddev->sysfs_action);
 		}
 		mddev_unlock(mddev);
 	}
@@ -6386,14 +6561,8 @@ static __exit void md_exit(void)
 	unregister_sysctl_table(raid_table_header);
 	remove_proc_entry("mdstat", NULL);
 	for_each_mddev(mddev, tmp) {
-		struct gendisk *disk = mddev->gendisk;
-		if (!disk)
-			continue;
 		export_array(mddev);
-		del_gendisk(disk);
-		put_disk(disk);
-		mddev->gendisk = NULL;
-		mddev_put(mddev);
+		mddev->hold_active = 0;
 	}
 }
 
@@ -6418,6 +6587,7 @@ static int set_ro(const char *val, struct kernel_param *kp)
 module_param_call(start_ro, set_ro, get_ro, NULL, S_IRUSR|S_IWUSR);
 module_param(start_dirty_degraded, int, S_IRUGO|S_IWUSR);
 
+module_param_call(new_array, add_named_array, NULL, NULL, S_IWUSR);
 
 EXPORT_SYMBOL(register_md_personality);
 EXPORT_SYMBOL(unregister_md_personality);
diff --git a/drivers/md/multipath.c b/drivers/md/multipath.c
index d4ac47d1127..f6d08f24167 100644
--- a/drivers/md/multipath.c
+++ b/drivers/md/multipath.c
@@ -408,7 +408,6 @@ static int multipath_run (mddev_t *mddev)
 	int disk_idx;
 	struct multipath_info *disk;
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 
 	if (mddev->level != LEVEL_MULTIPATH) {
 		printk("multipath: %s: raid level not set to multipath IO (%d)\n",
@@ -441,7 +440,7 @@ static int multipath_run (mddev_t *mddev)
 	}
 
 	conf->working_disks = 0;
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		disk_idx = rdev->raid_disk;
 		if (disk_idx < 0 ||
 		    disk_idx >= mddev->raid_disks)
diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index 8ac6488ad0d..c605ba80558 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -53,11 +53,10 @@ static int raid0_congested(void *data, int bits)
 static int create_strip_zones (mddev_t *mddev)
 {
 	int i, c, j;
-	sector_t current_offset, curr_zone_offset;
+	sector_t current_start, curr_zone_start;
 	sector_t min_spacing;
 	raid0_conf_t *conf = mddev_to_conf(mddev);
 	mdk_rdev_t *smallest, *rdev1, *rdev2, *rdev;
-	struct list_head *tmp1, *tmp2;
 	struct strip_zone *zone;
 	int cnt;
 	char b[BDEVNAME_SIZE];
@@ -67,19 +66,19 @@ static int create_strip_zones (mddev_t *mddev)
 	 */
 	conf->nr_strip_zones = 0;
  
-	rdev_for_each(rdev1, tmp1, mddev) {
-		printk("raid0: looking at %s\n",
+	list_for_each_entry(rdev1, &mddev->disks, same_set) {
+		printk(KERN_INFO "raid0: looking at %s\n",
 			bdevname(rdev1->bdev,b));
 		c = 0;
-		rdev_for_each(rdev2, tmp2, mddev) {
-			printk("raid0:   comparing %s(%llu)",
+		list_for_each_entry(rdev2, &mddev->disks, same_set) {
+			printk(KERN_INFO "raid0:   comparing %s(%llu)",
 			       bdevname(rdev1->bdev,b),
 			       (unsigned long long)rdev1->size);
-			printk(" with %s(%llu)\n",
+			printk(KERN_INFO " with %s(%llu)\n",
 			       bdevname(rdev2->bdev,b),
 			       (unsigned long long)rdev2->size);
 			if (rdev2 == rdev1) {
-				printk("raid0:   END\n");
+				printk(KERN_INFO "raid0:   END\n");
 				break;
 			}
 			if (rdev2->size == rdev1->size)
@@ -88,19 +87,20 @@ static int create_strip_zones (mddev_t *mddev)
 				 * Not unique, don't count it as a new
 				 * group
 				 */
-				printk("raid0:   EQUAL\n");
+				printk(KERN_INFO "raid0:   EQUAL\n");
 				c = 1;
 				break;
 			}
-			printk("raid0:   NOT EQUAL\n");
+			printk(KERN_INFO "raid0:   NOT EQUAL\n");
 		}
 		if (!c) {
-			printk("raid0:   ==> UNIQUE\n");
+			printk(KERN_INFO "raid0:   ==> UNIQUE\n");
 			conf->nr_strip_zones++;
-			printk("raid0: %d zones\n", conf->nr_strip_zones);
+			printk(KERN_INFO "raid0: %d zones\n",
+				conf->nr_strip_zones);
 		}
 	}
-	printk("raid0: FINAL %d zones\n", conf->nr_strip_zones);
+	printk(KERN_INFO "raid0: FINAL %d zones\n", conf->nr_strip_zones);
 
 	conf->strip_zone = kzalloc(sizeof(struct strip_zone)*
 				conf->nr_strip_zones, GFP_KERNEL);
@@ -119,16 +119,17 @@ static int create_strip_zones (mddev_t *mddev)
 	cnt = 0;
 	smallest = NULL;
 	zone->dev = conf->devlist;
-	rdev_for_each(rdev1, tmp1, mddev) {
+	list_for_each_entry(rdev1, &mddev->disks, same_set) {
 		int j = rdev1->raid_disk;
 
 		if (j < 0 || j >= mddev->raid_disks) {
-			printk("raid0: bad disk number %d - aborting!\n", j);
+			printk(KERN_ERR "raid0: bad disk number %d - "
+				"aborting!\n", j);
 			goto abort;
 		}
 		if (zone->dev[j]) {
-			printk("raid0: multiple devices for %d - aborting!\n",
-				j);
+			printk(KERN_ERR "raid0: multiple devices for %d - "
+				"aborting!\n", j);
 			goto abort;
 		}
 		zone->dev[j] = rdev1;
@@ -149,16 +150,16 @@ static int create_strip_zones (mddev_t *mddev)
 		cnt++;
 	}
 	if (cnt != mddev->raid_disks) {
-		printk("raid0: too few disks (%d of %d) - aborting!\n",
-			cnt, mddev->raid_disks);
+		printk(KERN_ERR "raid0: too few disks (%d of %d) - "
+			"aborting!\n", cnt, mddev->raid_disks);
 		goto abort;
 	}
 	zone->nb_dev = cnt;
-	zone->size = smallest->size * cnt;
-	zone->zone_offset = 0;
+	zone->sectors = smallest->size * cnt * 2;
+	zone->zone_start = 0;
 
-	current_offset = smallest->size;
-	curr_zone_offset = zone->size;
+	current_start = smallest->size * 2;
+	curr_zone_start = zone->sectors;
 
 	/* now do the other zones */
 	for (i = 1; i < conf->nr_strip_zones; i++)
@@ -166,40 +167,41 @@ static int create_strip_zones (mddev_t *mddev)
 		zone = conf->strip_zone + i;
 		zone->dev = conf->strip_zone[i-1].dev + mddev->raid_disks;
 
-		printk("raid0: zone %d\n", i);
-		zone->dev_offset = current_offset;
+		printk(KERN_INFO "raid0: zone %d\n", i);
+		zone->dev_start = current_start;
 		smallest = NULL;
 		c = 0;
 
 		for (j=0; j<cnt; j++) {
 			char b[BDEVNAME_SIZE];
 			rdev = conf->strip_zone[0].dev[j];
-			printk("raid0: checking %s ...", bdevname(rdev->bdev,b));
-			if (rdev->size > current_offset)
-			{
-				printk(" contained as device %d\n", c);
+			printk(KERN_INFO "raid0: checking %s ...",
+				bdevname(rdev->bdev, b));
+			if (rdev->size > current_start / 2) {
+				printk(KERN_INFO " contained as device %d\n",
+					c);
 				zone->dev[c] = rdev;
 				c++;
 				if (!smallest || (rdev->size <smallest->size)) {
 					smallest = rdev;
-					printk("  (%llu) is smallest!.\n", 
+					printk(KERN_INFO "  (%llu) is smallest!.\n",
 						(unsigned long long)rdev->size);
 				}
 			} else
-				printk(" nope.\n");
+				printk(KERN_INFO " nope.\n");
 		}
 
 		zone->nb_dev = c;
-		zone->size = (smallest->size - current_offset) * c;
-		printk("raid0: zone->nb_dev: %d, size: %llu\n",
-			zone->nb_dev, (unsigned long long)zone->size);
+		zone->sectors = (smallest->size * 2 - current_start) * c;
+		printk(KERN_INFO "raid0: zone->nb_dev: %d, sectors: %llu\n",
+			zone->nb_dev, (unsigned long long)zone->sectors);
 
-		zone->zone_offset = curr_zone_offset;
-		curr_zone_offset += zone->size;
+		zone->zone_start = curr_zone_start;
+		curr_zone_start += zone->sectors;
 
-		current_offset = smallest->size;
-		printk("raid0: current zone offset: %llu\n",
-			(unsigned long long)current_offset);
+		current_start = smallest->size * 2;
+		printk(KERN_INFO "raid0: current zone start: %llu\n",
+			(unsigned long long)current_start);
 	}
 
 	/* Now find appropriate hash spacing.
@@ -210,16 +212,16 @@ static int create_strip_zones (mddev_t *mddev)
 	 * strip though as it's size has no bearing on the efficacy of the hash
 	 * table.
 	 */
-	conf->hash_spacing = curr_zone_offset;
-	min_spacing = curr_zone_offset;
+	conf->spacing = curr_zone_start;
+	min_spacing = curr_zone_start;
 	sector_div(min_spacing, PAGE_SIZE/sizeof(struct strip_zone*));
 	for (i=0; i < conf->nr_strip_zones-1; i++) {
-		sector_t sz = 0;
-		for (j=i; j<conf->nr_strip_zones-1 &&
-			     sz < min_spacing ; j++)
-			sz += conf->strip_zone[j].size;
-		if (sz >= min_spacing && sz < conf->hash_spacing)
-			conf->hash_spacing = sz;
+		sector_t s = 0;
+		for (j = i; j < conf->nr_strip_zones - 1 &&
+				s < min_spacing; j++)
+			s += conf->strip_zone[j].sectors;
+		if (s >= min_spacing && s < conf->spacing)
+			conf->spacing = s;
 	}
 
 	mddev->queue->unplug_fn = raid0_unplug;
@@ -227,7 +229,7 @@ static int create_strip_zones (mddev_t *mddev)
 	mddev->queue->backing_dev_info.congested_fn = raid0_congested;
 	mddev->queue->backing_dev_info.congested_data = mddev;
 
-	printk("raid0: done.\n");
+	printk(KERN_INFO "raid0: done.\n");
 	return 0;
  abort:
 	return 1;
@@ -262,10 +264,9 @@ static int raid0_mergeable_bvec(struct request_queue *q,
 static int raid0_run (mddev_t *mddev)
 {
 	unsigned  cur=0, i=0, nb_zone;
-	s64 size;
+	s64 sectors;
 	raid0_conf_t *conf;
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 
 	if (mddev->chunk_size == 0) {
 		printk(KERN_ERR "md/raid0: non-zero chunk size required.\n");
@@ -291,54 +292,54 @@ static int raid0_run (mddev_t *mddev)
 
 	/* calculate array device size */
 	mddev->array_sectors = 0;
-	rdev_for_each(rdev, tmp, mddev)
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		mddev->array_sectors += rdev->size * 2;
 
-	printk("raid0 : md_size is %llu blocks.\n", 
-		(unsigned long long)mddev->array_sectors / 2);
-	printk("raid0 : conf->hash_spacing is %llu blocks.\n",
-		(unsigned long long)conf->hash_spacing);
+	printk(KERN_INFO "raid0 : md_size is %llu sectors.\n",
+		(unsigned long long)mddev->array_sectors);
+	printk(KERN_INFO "raid0 : conf->spacing is %llu sectors.\n",
+		(unsigned long long)conf->spacing);
 	{
-		sector_t s = mddev->array_sectors / 2;
-		sector_t space = conf->hash_spacing;
+		sector_t s = mddev->array_sectors;
+		sector_t space = conf->spacing;
 		int round;
-		conf->preshift = 0;
+		conf->sector_shift = 0;
 		if (sizeof(sector_t) > sizeof(u32)) {
 			/*shift down space and s so that sector_div will work */
 			while (space > (sector_t) (~(u32)0)) {
 				s >>= 1;
 				space >>= 1;
 				s += 1; /* force round-up */
-				conf->preshift++;
+				conf->sector_shift++;
 			}
 		}
 		round = sector_div(s, (u32)space) ? 1 : 0;
 		nb_zone = s + round;
 	}
-	printk("raid0 : nb_zone is %d.\n", nb_zone);
+	printk(KERN_INFO "raid0 : nb_zone is %d.\n", nb_zone);
 
-	printk("raid0 : Allocating %Zd bytes for hash.\n",
+	printk(KERN_INFO "raid0 : Allocating %zu bytes for hash.\n",
 				nb_zone*sizeof(struct strip_zone*));
 	conf->hash_table = kmalloc (sizeof (struct strip_zone *)*nb_zone, GFP_KERNEL);
 	if (!conf->hash_table)
 		goto out_free_conf;
-	size = conf->strip_zone[cur].size;
+	sectors = conf->strip_zone[cur].sectors;
 
 	conf->hash_table[0] = conf->strip_zone + cur;
 	for (i=1; i< nb_zone; i++) {
-		while (size <= conf->hash_spacing) {
+		while (sectors <= conf->spacing) {
 			cur++;
-			size += conf->strip_zone[cur].size;
+			sectors += conf->strip_zone[cur].sectors;
 		}
-		size -= conf->hash_spacing;
+		sectors -= conf->spacing;
 		conf->hash_table[i] = conf->strip_zone + cur;
 	}
-	if (conf->preshift) {
-		conf->hash_spacing >>= conf->preshift;
-		/* round hash_spacing up so when we divide by it, we
+	if (conf->sector_shift) {
+		conf->spacing >>= conf->sector_shift;
+		/* round spacing up so when we divide by it, we
 		 * err on the side of too-low, which is safest
 		 */
-		conf->hash_spacing++;
+		conf->spacing++;
 	}
 
 	/* calculate the max read-ahead size.
@@ -387,12 +388,12 @@ static int raid0_stop (mddev_t *mddev)
 static int raid0_make_request (struct request_queue *q, struct bio *bio)
 {
 	mddev_t *mddev = q->queuedata;
-	unsigned int sect_in_chunk, chunksize_bits,  chunk_size, chunk_sects;
+	unsigned int sect_in_chunk, chunksect_bits, chunk_sects;
 	raid0_conf_t *conf = mddev_to_conf(mddev);
 	struct strip_zone *zone;
 	mdk_rdev_t *tmp_dev;
 	sector_t chunk;
-	sector_t block, rsect;
+	sector_t sector, rsect;
 	const int rw = bio_data_dir(bio);
 	int cpu;
 
@@ -407,11 +408,9 @@ static int raid0_make_request (struct request_queue *q, struct bio *bio)
 		      bio_sectors(bio));
 	part_stat_unlock();
 
-	chunk_size = mddev->chunk_size >> 10;
 	chunk_sects = mddev->chunk_size >> 9;
-	chunksize_bits = ffz(~chunk_size);
-	block = bio->bi_sector >> 1;
-	
+	chunksect_bits = ffz(~chunk_sects);
+	sector = bio->bi_sector;
 
 	if (unlikely(chunk_sects < (bio->bi_sector & (chunk_sects - 1)) + (bio->bi_size >> 9))) {
 		struct bio_pair *bp;
@@ -434,28 +433,27 @@ static int raid0_make_request (struct request_queue *q, struct bio *bio)
  
 
 	{
-		sector_t x = block >> conf->preshift;
-		sector_div(x, (u32)conf->hash_spacing);
+		sector_t x = sector >> conf->sector_shift;
+		sector_div(x, (u32)conf->spacing);
 		zone = conf->hash_table[x];
 	}
- 
-	while (block >= (zone->zone_offset + zone->size)) 
+
+	while (sector >= zone->zone_start + zone->sectors)
 		zone++;
-    
-	sect_in_chunk = bio->bi_sector & ((chunk_size<<1) -1);
+
+	sect_in_chunk = bio->bi_sector & (chunk_sects - 1);
 
 
 	{
-		sector_t x =  (block - zone->zone_offset) >> chunksize_bits;
+		sector_t x = (sector - zone->zone_start) >> chunksect_bits;
 
 		sector_div(x, zone->nb_dev);
 		chunk = x;
 
-		x = block >> chunksize_bits;
+		x = sector >> chunksect_bits;
 		tmp_dev = zone->dev[sector_div(x, zone->nb_dev)];
 	}
-	rsect = (((chunk << chunksize_bits) + zone->dev_offset)<<1)
-		+ sect_in_chunk;
+	rsect = (chunk << chunksect_bits) + zone->dev_start + sect_in_chunk;
  
 	bio->bi_bdev = tmp_dev->bdev;
 	bio->bi_sector = rsect + tmp_dev->data_offset;
@@ -467,7 +465,7 @@ static int raid0_make_request (struct request_queue *q, struct bio *bio)
 
 bad_map:
 	printk("raid0_make_request bug: can't convert block across chunks"
-		" or bigger than %dk %llu %d\n", chunk_size, 
+		" or bigger than %dk %llu %d\n", chunk_sects / 2,
 		(unsigned long long)bio->bi_sector, bio->bi_size >> 10);
 
 	bio_io_error(bio);
@@ -492,10 +490,10 @@ static void raid0_status (struct seq_file *seq, mddev_t *mddev)
 			seq_printf(seq, "%s/", bdevname(
 				conf->strip_zone[j].dev[k]->bdev,b));
 
-		seq_printf(seq, "] zo=%d do=%d s=%d\n",
-				conf->strip_zone[j].zone_offset,
-				conf->strip_zone[j].dev_offset,
-				conf->strip_zone[j].size);
+		seq_printf(seq, "] zs=%d ds=%d s=%d\n",
+				conf->strip_zone[j].zone_start,
+				conf->strip_zone[j].dev_start,
+				conf->strip_zone[j].sectors);
 	}
 #endif
 	seq_printf(seq, " %dk chunks", mddev->chunk_size/1024);
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index 9c788e2489b..7b4f5f7155d 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -1016,12 +1016,16 @@ static void error(mddev_t *mddev, mdk_rdev_t *rdev)
 	 * else mark the drive as failed
 	 */
 	if (test_bit(In_sync, &rdev->flags)
-	    && (conf->raid_disks - mddev->degraded) == 1)
+	    && (conf->raid_disks - mddev->degraded) == 1) {
 		/*
 		 * Don't fail the drive, act as though we were just a
-		 * normal single drive
+		 * normal single drive.
+		 * However don't try a recovery from this drive as
+		 * it is very likely to fail.
 		 */
+		mddev->recovery_disabled = 1;
 		return;
+	}
 	if (test_and_clear_bit(In_sync, &rdev->flags)) {
 		unsigned long flags;
 		spin_lock_irqsave(&conf->device_lock, flags);
@@ -1919,7 +1923,6 @@ static int run(mddev_t *mddev)
 	int i, j, disk_idx;
 	mirror_info_t *disk;
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 
 	if (mddev->level != 1) {
 		printk("raid1: %s: raid level not set to mirroring (%d)\n",
@@ -1964,7 +1967,7 @@ static int run(mddev_t *mddev)
 	spin_lock_init(&conf->device_lock);
 	mddev->queue->queue_lock = &conf->device_lock;
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		disk_idx = rdev->raid_disk;
 		if (disk_idx >= mddev->raid_disks
 		    || disk_idx < 0)
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 970a96ef9b1..6736d6dff98 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -2025,7 +2025,6 @@ static int run(mddev_t *mddev)
 	int i, disk_idx;
 	mirror_info_t *disk;
 	mdk_rdev_t *rdev;
-	struct list_head *tmp;
 	int nc, fc, fo;
 	sector_t stride, size;
 
@@ -2108,7 +2107,7 @@ static int run(mddev_t *mddev)
 	spin_lock_init(&conf->device_lock);
 	mddev->queue->queue_lock = &conf->device_lock;
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		disk_idx = rdev->raid_disk;
 		if (disk_idx >= mddev->raid_disks
 		    || disk_idx < 0)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index a36a7435edf..a5ba080d303 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -3998,7 +3998,6 @@ static int run(mddev_t *mddev)
 	int raid_disk, memory;
 	mdk_rdev_t *rdev;
 	struct disk_info *disk;
-	struct list_head *tmp;
 	int working_disks = 0;
 
 	if (mddev->level != 5 && mddev->level != 4 && mddev->level != 6) {
@@ -4108,7 +4107,7 @@ static int run(mddev_t *mddev)
 
 	pr_debug("raid5: run(%s) called.\n", mdname(mddev));
 
-	rdev_for_each(rdev, tmp, mddev) {
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		raid_disk = rdev->raid_disk;
 		if (raid_disk >= conf->raid_disks
 		    || raid_disk < 0)
@@ -4533,7 +4532,6 @@ static int raid5_start_reshape(mddev_t *mddev)
 {
 	raid5_conf_t *conf = mddev_to_conf(mddev);
 	mdk_rdev_t *rdev;
-	struct list_head *rtmp;
 	int spares = 0;
 	int added_devices = 0;
 	unsigned long flags;
@@ -4541,7 +4539,7 @@ static int raid5_start_reshape(mddev_t *mddev)
 	if (test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
 		return -EBUSY;
 
-	rdev_for_each(rdev, rtmp, mddev)
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		if (rdev->raid_disk < 0 &&
 		    !test_bit(Faulty, &rdev->flags))
 			spares++;
@@ -4563,7 +4561,7 @@ static int raid5_start_reshape(mddev_t *mddev)
 	/* Add some new drives, as many as will fit.
 	 * We know there are enough to make the newly sized array work.
 	 */
-	rdev_for_each(rdev, rtmp, mddev)
+	list_for_each_entry(rdev, &mddev->disks, same_set)
 		if (rdev->raid_disk < 0 &&
 		    !test_bit(Faulty, &rdev->flags)) {
 			if (raid5_add_disk(mddev, rdev) == 0) {